new api

Co-authored-by: yosagi <11404771+yosagi@users.noreply.github.com>"

.gitignore

@@ -54,21 +54,6 @@ coverage.xml
 # Translations
 *.mo
 *.pot
-# Django stuff:
-*.log
-local_settings.py
-db.sqlite3
-db.sqlite3-journal
-
-# Flask stuff:
-instance/
-.webassets-cache
-
-# Scrapy stuff:
-.scrapy
-
-# Sphinx documentation
-docs/_build/
 
 # PyBuilder
 target/
@@ -138,4 +123,5 @@ test_*.py
 launch.json
 .DS_Store
 test/*
-nllb-200-distilled-600M-ctranslate2/*
+nllb-200-distilled-600M-ctranslate2/*
+*.mp3

DEV_NOTES.md

@@ -18,8 +18,29 @@ Decoder weights: 59110771 bytes
 Encoder weights: 15268874 bytes
 
 
+# 2. Translation: Faster model for each system
 
-# 2. SortFormer Diarization: 4-to-2 Speaker Constraint Algorithm
+## Benchmark Results
+
+Testing on MacBook M3 with NLLB-200-distilled-600M model:
+
+### Standard Transformers vs CTranslate2
+
+| Test Text | Standard Inference Time | CTranslate2 Inference Time | Speedup |
+|-----------|-------------------------|---------------------------|---------|
+| UN Chief says there is no military solution in Syria | 0.9395s | 2.0472s | 0.5x |
+| The rapid advancement of AI technology is transforming various industries | 0.7171s | 1.7516s | 0.4x |
+| Climate change poses a significant threat to global ecosystems | 0.8533s | 1.8323s | 0.5x |
+| International cooperation is essential for addressing global challenges | 0.7209s | 1.3575s | 0.5x |
+| The development of renewable energy sources is crucial for a sustainable future | 0.8760s | 1.5589s | 0.6x |
+
+**Results:**
+- Total Standard time: 4.1068s
+- Total CTranslate2 time: 8.5476s
+- CTranslate2 is slower on this system --> Use Transformers, and ideally we would have an mlx implementation.
+
+
+# 3. SortFormer Diarization: 4-to-2 Speaker Constraint Algorithm
 
 Transform a diarization model that predicts up to 4 speakers into one that predicts up to 2 speakers by mapping the output predictions.
 
@@ -67,4 +88,4 @@ ELSE:
     AS_2 ← B
 
 to finish
-```
+```

LICENSE

@@ -1,52 +1,210 @@
-# License
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
 
-## Main Software License
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
 
-MIT License
+   1. Definitions.
 
-Copyright (c) 2025 Quentin Fuxa.  
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
 
-Permission is hereby granted, free of charge, to any person obtaining a copy  
-of this software and associated documentation files (the "Software"), to deal  
-in the Software without restriction, including without limitation the rights  
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell  
-copies of the Software, and to permit persons to whom the Software is  
-furnished to do so, subject to the following conditions:
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
 
-The above copyright notice and this permission notice shall be included in all  
-copies or substantial portions of the Software.
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
 
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR  
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,  
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE  
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER  
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,  
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE  
-SOFTWARE.
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
 
-## SimulStreaming Backend License
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
 
-**When using the SimulStreaming backend (SimulWhisper), additional licensing terms apply:**
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
 
-SimulStreaming (https://github.com/ufal/SimulStreaming) is dual-licensed:
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
 
-### 🔹 Non-Commercial Use
-You may use SimulStreaming under the **PolyForm Noncommercial License 1.0.0** if you obtain the code through the GitHub repository. This license is **free of charge** and comes with **no obligations** for non-commercial users.
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
 
-### 🔸 Commercial Use
-Understanding who uses SimulStreaming commercially helps improve and prioritize development. Therefore, **registration is required** for those who acquire a commercial license.
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
 
-Commercial licenses are planned to be **affordable** to SMEs and individuals. They are considering providing commercial licenses either for free or for a symbolic one-time fee, and may also provide additional support. You can share your preference via the [questionnaire](https://forms.cloud.microsoft.com/e/7tCxb4gJfB).
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
 
-You can also leave your contact [there](https://forms.cloud.microsoft.com/e/7tCxb4gJfB) to be notified when commercial licenses become available.
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
 
-**Contact for SimulStreaming licensing:**  
-[Dominik Macháček](https://ufal.mff.cuni.cz/dominik-machacek/), machacek@ufal.mff.cuni.cz
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
 
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2025 Quentin Fuxa
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.
 ---
 
 ## Based on:
-- **whisper_streaming** by ÚFAL – MIT License – https://github.com/ufal/whisper_streaming. The original work by ÚFAL. License: https://github.com/ufal/whisper_streaming/blob/main/LICENSE  
-- **silero-vad** by Snakers4 – MIT License – https://github.com/snakers4/silero-vad. The work by Snakers4 (silero-vad). License: https://github.com/snakers4/silero-vad/blob/f6b1294cb27590fb2452899df98fb234dfef1134/LICENSE  
-- **Diart** by juanmc2005 – MIT License – https://github.com/juanmc2005/diart. The work in Diart by juanmc2005. License: https://github.com/juanmc2005/diart/blob/main/LICENSE
-- **SimulStreaming** by ÚFAL – Dual License (PolyForm Noncommercial License 1.0.0 / Commercial License) – https://github.com/ufal/SimulStreaming
+- **SimulWhisper** by Speech and Audio Technology LAB of Tsinghua University – Apache-2.0 – https://github.com/ufal/SimulStreaming
+- **SimulStreaming** by ÚFAL – MIT License – https://github.com/ufal/SimulStreaming
+- **NeMo** by NVidia - Apache-2.0 - https://github.com/NVIDIA-NeMo/NeMo 
+- **whisper_streaming** by ÚFAL – MIT License – https://github.com/ufal/whisper_streaming.
+- **silero-vad** by Snakers4 – MIT License – https://github.com/snakers4/silero-vad.
+- **Diart** by juanmc2005 – MIT License – https://github.com/juanmc2005/diart.

README.md

@@ -10,17 +10,17 @@
 <a href="https://pypi.org/project/whisperlivekit/"><img alt="PyPI Version" src="https://img.shields.io/pypi/v/whisperlivekit?color=g"></a>
 <a href="https://pepy.tech/project/whisperlivekit"><img alt="PyPI Downloads" src="https://static.pepy.tech/personalized-badge/whisperlivekit?period=total&units=international_system&left_color=grey&right_color=brightgreen&left_text=installations"></a>
 <a href="https://pypi.org/project/whisperlivekit/"><img alt="Python Versions" src="https://img.shields.io/badge/python-3.9--3.15-dark_green"></a>
-<a href="https://github.com/QuentinFuxa/WhisperLiveKit/blob/main/LICENSE"><img alt="License" src="https://img.shields.io/badge/License-MIT/Dual Licensed-dark_green"></a>
+<a href="https://github.com/QuentinFuxa/WhisperLiveKit/blob/main/LICENSE"><img alt="License" src="https://img.shields.io/badge/License-Apache 2.0-dark_green"></a>
 </p>
 
 
-Real-time speech transcription directly to your browser, with a ready-to-use backend+server and a simple frontend. ✨
+Real-time transcription directly to your browser, with a ready-to-use backend+server and a simple frontend.
 
 #### Powered by Leading Research:
 
-- [SimulStreaming](https://github.com/ufal/SimulStreaming) (SOTA 2025) - Ultra-low latency transcription with AlignAtt policy
-- [NLLB](https://arxiv.org/abs/2207.04672), ([distilled](https://huggingface.co/entai2965/nllb-200-distilled-600M-ctranslate2)) (2024) - Translation to more than 100 languages.
-- [WhisperStreaming](https://github.com/ufal/whisper_streaming) (SOTA 2023) - Low latency transcription with LocalAgreement policy
+- Simul-[Whisper](https://github.com/backspacetg/simul_whisper)/[Streaming](https://github.com/ufal/SimulStreaming) (SOTA 2025) - Ultra-low latency transcription using [AlignAtt policy](https://arxiv.org/pdf/2305.11408)
+- [NLLW](https://github.com/QuentinFuxa/NoLanguageLeftWaiting) (2025), based on [distilled](https://huggingface.co/entai2965/nllb-200-distilled-600M-ctranslate2) [NLLB](https://arxiv.org/abs/2207.04672) (2022, 2024) - Simulatenous translation from & to 200 languages.
+- [WhisperStreaming](https://github.com/ufal/whisper_streaming) (SOTA 2023) - Low latency transcription using [LocalAgreement policy](https://www.isca-archive.org/interspeech_2020/liu20s_interspeech.pdf)
 - [Streaming Sortformer](https://arxiv.org/abs/2507.18446) (SOTA 2025) - Advanced real-time speaker diarization
 - [Diart](https://github.com/juanmc2005/diart) (SOTA 2021) - Real-time speaker diarization
 - [Silero VAD](https://github.com/snakers4/silero-vad) (2024) - Enterprise-grade Voice Activity Detection
@@ -42,15 +42,6 @@ pip install whisperlivekit
 ```
 > You can also clone the repo and `pip install -e .` for the latest version.
 
-
->  **FFmpeg is required** and must be installed before using WhisperLiveKit
-> 
-> | OS | How to install |
-> |-----------|-------------|
->  | Ubuntu/Debian | `sudo apt install ffmpeg` |
-> | MacOS | `brew install ffmpeg` |
-> | Windows | Download .exe from https://ffmpeg.org/download.html and add to PATH |
-
 #### Quick Start
 1. **Start the transcription server:**
    ```bash
@@ -63,15 +54,23 @@ pip install whisperlivekit
 > - See [tokenizer.py](https://github.com/QuentinFuxa/WhisperLiveKit/blob/main/whisperlivekit/simul_whisper/whisper/tokenizer.py) for the list of all available languages.
 > - For HTTPS requirements, see the **Parameters** section for SSL configuration options.
 
- 
+#### Use it to capture audio from web pages.
+
+Go to `chrome-extension` for instructions.
+
+<p align="center">
+<img src="https://raw.githubusercontent.com/QuentinFuxa/WhisperLiveKit/refs/heads/main/chrome-extension/demo-extension.png" alt="WhisperLiveKit Demo" width="600">
+</p>
+
+
 
 #### Optional Dependencies
 
 | Optional | `pip install` |
 |-----------|-------------|
-| **Speaker diarization with Sortformer** | `git+https://github.com/NVIDIA/NeMo.git@main#egg=nemo_toolkit[asr]` |
-| **Apple Silicon optimized backend** | `mlx-whisper` |
-| **NLLB Translation** | `huggingface_hub` & `transformers` |
+| **Speaker diarization** | `git+https://github.com/NVIDIA/NeMo.git@main#egg=nemo_toolkit[asr]` |
+| **Apple Silicon optimizations** | `mlx-whisper` |
+| **Translation** | `nllw` |
 | *[Not recommanded]*  Speaker diarization with Diart | `diart` |
 | *[Not recommanded]*  Original Whisper backend | `whisper` |
 | *[Not recommanded]*  Improved timestamps backend | `whisper-timestamped` |
@@ -86,11 +85,11 @@ See  **Parameters & Configuration** below on how to use them.
 **Command-line Interface**: Start the transcription server with various options:
 
 ```bash
-# Use better model than default (small)
-whisperlivekit-server --model large-v3
+# Large model and translate from french to danish
+whisperlivekit-server --model large-v3 --language fr --target-language da
 
-# Advanced configuration with diarization and language
-whisperlivekit-server --host 0.0.0.0 --port 8000 --model medium --diarization --language fr
+# Diarization and server listening on */80 
+whisperlivekit-server --host 0.0.0.0 --port 80 --model medium --diarization --language fr
 ```
 
 
@@ -137,26 +136,16 @@ async def websocket_endpoint(websocket: WebSocket):
 
 ## Parameters & Configuration
 
-An important list of parameters can be changed. But what *should* you change?
-- the `--model` size. List and recommandations [here](https://github.com/QuentinFuxa/WhisperLiveKit/blob/main/available_models.md)
-- the `--language`.  List [here](https://github.com/QuentinFuxa/WhisperLiveKit/blob/main/whisperlivekit/simul_whisper/whisper/tokenizer.py). If you use `auto`, the model attempts to detect the language automatically, but it tends to bias towards English.
-- the `--backend` ? you can switch to `--backend faster-whisper` if  `simulstreaming` does not work correctly or if you prefer to avoid the dual-license requirements.
-- `--warmup-file`, if you have one
-- `--task translate`, to translate in english
-- `--host`, `--port`, `--ssl-certfile`, `--ssl-keyfile`, if you set up a server
-- `--diarization`, if you want to use it.
-- [BETA] `--target-language`, to translate using NLLB. [118 languages available](https://github.com/QuentinFuxa/WhisperLiveKit/blob/main/whisperlivekit/translation/mapping_languages.py). If you want to translate to english, you should rather use `--task translate`, since Whisper can do it directly.
-
-### Full list of parameters :
 
 | Parameter | Description | Default |
 |-----------|-------------|---------|
-| `--model` | Whisper model size. | `small` |
-| `--language` | Source language code or `auto` | `auto` |
-| `--task` | Set to `translate` to translate to english | `transcribe` |
-| `--target-language` | [BETA] Translation language target. Ex: `fr` | `None` |
-| `--backend` | Processing backend | `simulstreaming` |
-| `--min-chunk-size` | Minimum audio chunk size (seconds) | `1.0` |
+| `--model` | Whisper model size. List and recommandations [here](https://github.com/QuentinFuxa/WhisperLiveKit/blob/main/docs/available_models.md) | `small` |
+| `--model-path` | .pt file/directory containing whisper model. Overrides `--model`. Recommandations [here](https://github.com/QuentinFuxa/WhisperLiveKit/blob/main/docs/models_compatible_formats.md) | `None` |
+| `--language` | List [here](https://github.com/QuentinFuxa/WhisperLiveKit/blob/main/whisperlivekit/simul_whisper/whisper/tokenizer.py). If you use `auto`, the model attempts to detect the language automatically, but it tends to bias towards English. | `auto` |
+| `--target-language` | If sets, translate to using NLLB. Ex: `fr`. [200 languages available](https://github.com/QuentinFuxa/WhisperLiveKit/blob/main/docs/supported_languages.md). If you want to translate to english, you should rather use `--task translate`, since Whisper can do it directly. | `None` |
+| `--task` | Set to `translate` to translate *only* to english, using Whisper translation. | `transcribe` |
+| `--diarization` | Enable speaker identification | `False` |
+| `--backend` | Processing backend. You can switch to `faster-whisper` if  `simulstreaming` does not work correctly | `simulstreaming` |
 | `--no-vac` | Disable Voice Activity Controller | `False` |
 | `--no-vad` | Disable Voice Activity Detection | `False` |
 | `--warmup-file` | Audio file path for model warmup | `jfk.wav` |
@@ -164,12 +153,25 @@ An important list of parameters can be changed. But what *should* you change?
 | `--port` | Server port | `8000` |
 | `--ssl-certfile` | Path to the SSL certificate file (for HTTPS support) | `None` |
 | `--ssl-keyfile` | Path to the SSL private key file (for HTTPS support) | `None` |
-| `--pcm-input` | raw PCM (s16le) data is expected as input and FFmpeg will be bypassed. | `False` |
+| `--forwarded-allow-ips` | Ip or Ips allowed to reverse proxy the whisperlivekit-server. Supported types are  IP Addresses (e.g. 127.0.0.1), IP Networks (e.g. 10.100.0.0/16), or Literals (e.g. /path/to/socket.sock) | `None` |
+| `--pcm-input` | raw PCM (s16le) data is expected as input and FFmpeg will be bypassed. Frontend will use AudioWorklet instead of MediaRecorder | `False` |
 
+| Translation options | Description | Default |
+|-----------|-------------|---------|
+| `--nllb-backend` | `transformers` or `ctranslate2` | `ctranslate2` |
+| `--nllb-size` | `600M` or `1.3B` | `600M` |
+
+| Diarization options | Description | Default |
+|-----------|-------------|---------|
+| `--diarization-backend` |  `diart` or `sortformer` | `sortformer` |
+| `--disable-punctuation-split` |  Disable punctuation based splits. See #214 | `False` |
+| `--segmentation-model` | Hugging Face model ID for Diart segmentation model. [Available models](https://github.com/juanmc2005/diart/tree/main?tab=readme-ov-file#pre-trained-models) | `pyannote/segmentation-3.0` |
+| `--embedding-model` | Hugging Face model ID for Diart embedding model. [Available models](https://github.com/juanmc2005/diart/tree/main?tab=readme-ov-file#pre-trained-models) | `speechbrain/spkrec-ecapa-voxceleb` |
 
 | SimulStreaming backend options | Description | Default |
 |-----------|-------------|---------|
 | `--disable-fast-encoder` | Disable Faster Whisper or MLX Whisper backends for the encoder (if installed). Inference can be slower but helpful when GPU memory is limited | `False` |
+| `--custom-alignment-heads` | Use your own alignment heads, useful when `--model-dir` is used | `None` |
 | `--frame-threshold` | AlignAtt frame threshold (lower = faster, higher = more accurate) | `25` |
 | `--beams` | Number of beams for beam search (1 = greedy decoding) | `1` |
 | `--decoder` | Force decoder type (`beam` or `greedy`) | `auto` |
@@ -180,29 +182,19 @@ An important list of parameters can be changed. But what *should* you change?
 | `--init-prompt` | Initial prompt for the model | `None` |
 | `--static-init-prompt` | Static prompt that doesn't scroll | `None` |
 | `--max-context-tokens` | Maximum context tokens | `None` |
-| `--model-path` | Direct path to .pt model file. Download it if not found | `./base.pt` |
 | `--preload-model-count` | Optional. Number of models to preload in memory to speed up loading (set up to the expected number of concurrent users) | `1` |
 
 
+
 | WhisperStreaming backend options | Description | Default |
 |-----------|-------------|---------|
 | `--confidence-validation` | Use confidence scores for faster validation | `False` |
 | `--buffer_trimming` | Buffer trimming strategy (`sentence` or `segment`) | `segment` |
 
-| Diarization options | Description | Default |
-|-----------|-------------|---------|
-| `--diarization` | Enable speaker identification | `False` |
-| `--diarization-backend` |  `diart` or `sortformer` | `sortformer` |
-| `--disable-punctuation-split` |  Disable punctuation based splits. See #214 | `False` |
-| `--segmentation-model` | Hugging Face model ID for Diart segmentation model. [Available models](https://github.com/juanmc2005/diart/tree/main?tab=readme-ov-file#pre-trained-models) | `pyannote/segmentation-3.0` |
-| `--embedding-model` | Hugging Face model ID for Diart embedding model. [Available models](https://github.com/juanmc2005/diart/tree/main?tab=readme-ov-file#pre-trained-models) | `speechbrain/spkrec-ecapa-voxceleb` |
 
 
-> For diarization using Diart, you need access to pyannote.audio models:
-> 1. [Accept user conditions](https://huggingface.co/pyannote/segmentation) for the `pyannote/segmentation` model
-> 2. [Accept user conditions](https://huggingface.co/pyannote/segmentation-3.0) for the `pyannote/segmentation-3.0` model
-> 3. [Accept user conditions](https://huggingface.co/pyannote/embedding) for the `pyannote/embedding` model
->4. Login with HuggingFace: `huggingface-cli login`
+
+> For diarization using Diart, you need to accept user conditions [here](https://huggingface.co/pyannote/segmentation) for the `pyannote/segmentation` model, [here](https://huggingface.co/pyannote/segmentation-3.0) for the `pyannote/segmentation-3.0` model and [here](https://huggingface.co/pyannote/embedding) for the `pyannote/embedding` model. **Then**, login to HuggingFace: `huggingface-cli login`
 
 ### 🚀 Deployment Guide
 

chrome-extension/README.md

@@ -1,11 +1,13 @@
-## WhisperLiveKit Chrome Extension v0.1.0
-Capture the audio of your current tab, transcribe or translate it using WhisperliveKit. **Still unstable**
+## WhisperLiveKit Chrome Extension v0.1.1
+Capture the audio of your current tab, transcribe diarize and translate it using WhisperliveKit, in Chrome and other Chromium-based browsers.
+
+> Currently, only the tab audio is captured; your microphone audio is not recorded.
 
 <img src="https://raw.githubusercontent.com/QuentinFuxa/WhisperLiveKit/refs/heads/main/chrome-extension/demo-extension.png" alt="WhisperLiveKit Demo" width="730">
 
 ## Running this extension
-1. Clone this repository.
-2. Load this directory in Chrome as an unpacked extension.
+1. Run `python sync_extension.py` to copy frontend files to the `chrome-extension` directory.
+2. Load the `chrome-extension` directory in Chrome as an unpacked extension.
 
 
 ## Devs:

chrome-extension/live_transcription.js

@@ -1,669 +0,0 @@
-/* Theme, WebSocket, recording, rendering logic extracted from inline script and adapted for segmented theme control and WS caption */
-let isRecording = false;
-let websocket = null;
-let recorder = null;
-let chunkDuration = 100;
-let websocketUrl = "ws://localhost:8000/asr";
-let userClosing = false;
-let wakeLock = null;
-let startTime = null;
-let timerInterval = null;
-let audioContext = null;
-let analyser = null;
-let microphone = null;
-let waveCanvas = document.getElementById("waveCanvas");
-let waveCtx = waveCanvas.getContext("2d");
-let animationFrame = null;
-let waitingForStop = false;
-let lastReceivedData = null;
-let lastSignature = null;
-let availableMicrophones = [];
-let selectedMicrophoneId = null;
-
-waveCanvas.width = 60 * (window.devicePixelRatio || 1);
-waveCanvas.height = 30 * (window.devicePixelRatio || 1);
-waveCtx.scale(window.devicePixelRatio || 1, window.devicePixelRatio || 1);
-
-const statusText = document.getElementById("status");
-const recordButton = document.getElementById("recordButton");
-const chunkSelector = document.getElementById("chunkSelector");
-const websocketInput = document.getElementById("websocketInput");
-const websocketDefaultSpan = document.getElementById("wsDefaultUrl");
-const linesTranscriptDiv = document.getElementById("linesTranscript");
-const timerElement = document.querySelector(".timer");
-const themeRadios = document.querySelectorAll('input[name="theme"]');
-const microphoneSelect = document.getElementById("microphoneSelect");
-const settingsToggle = document.getElementById("settingsToggle");
-const settingsDiv = document.querySelector(".settings");
-
-
-
-chrome.runtime.onInstalled.addListener((details) => {
-    if (details.reason.search(/install/g) === -1) {
-        return
-    }
-    chrome.tabs.create({
-        url: chrome.runtime.getURL("welcome.html"),
-        active: true
-    })
-})
-
-function getWaveStroke() {
-  const styles = getComputedStyle(document.documentElement);
-  const v = styles.getPropertyValue("--wave-stroke").trim();
-  return v || "#000";
-}
-
-let waveStroke = getWaveStroke();
-function updateWaveStroke() {
-  waveStroke = getWaveStroke();
-}
-
-function applyTheme(pref) {
-  if (pref === "light") {
-    document.documentElement.setAttribute("data-theme", "light");
-  } else if (pref === "dark") {
-    document.documentElement.setAttribute("data-theme", "dark");
-  } else {
-    document.documentElement.removeAttribute("data-theme");
-  }
-  updateWaveStroke();
-}
-
-// Persisted theme preference
-const savedThemePref = localStorage.getItem("themePreference") || "system";
-applyTheme(savedThemePref);
-if (themeRadios.length) {
-  themeRadios.forEach((r) => {
-    r.checked = r.value === savedThemePref;
-    r.addEventListener("change", () => {
-      if (r.checked) {
-        localStorage.setItem("themePreference", r.value);
-        applyTheme(r.value);
-      }
-    });
-  });
-}
-
-// React to OS theme changes when in "system" mode
-const darkMq = window.matchMedia && window.matchMedia("(prefers-color-scheme: dark)");
-const handleOsThemeChange = () => {
-  const pref = localStorage.getItem("themePreference") || "system";
-  if (pref === "system") updateWaveStroke();
-};
-if (darkMq && darkMq.addEventListener) {
-  darkMq.addEventListener("change", handleOsThemeChange);
-} else if (darkMq && darkMq.addListener) {
-  // deprecated, but included for Safari compatibility
-  darkMq.addListener(handleOsThemeChange);
-}
-
-async function enumerateMicrophones() {
-  try {
-      const micPermission = await navigator.permissions.query({
-    name: "microphone",
-  });
-  
-    const stream = await navigator.mediaDevices.getUserMedia({ audio: true });
-    stream.getTracks().forEach(track => track.stop());
-
-    const devices = await navigator.mediaDevices.enumerateDevices();
-    availableMicrophones = devices.filter(device => device.kind === 'audioinput');
-
-    populateMicrophoneSelect();
-    console.log(`Found ${availableMicrophones.length} microphone(s)`);
-  } catch (error) {
-    console.error('Error enumerating microphones:', error);
-    statusText.textContent = "Error accessing microphones. Please grant permission.";
-  }
-}
-
-function populateMicrophoneSelect() {
-  if (!microphoneSelect) return;
-
-  microphoneSelect.innerHTML = '<option value="">Default Microphone</option>';
-
-  availableMicrophones.forEach((device, index) => {
-    const option = document.createElement('option');
-    option.value = device.deviceId;
-    option.textContent = device.label || `Microphone ${index + 1}`;
-    microphoneSelect.appendChild(option);
-  });
-
-  const savedMicId = localStorage.getItem('selectedMicrophone');
-  if (savedMicId && availableMicrophones.some(mic => mic.deviceId === savedMicId)) {
-    microphoneSelect.value = savedMicId;
-    selectedMicrophoneId = savedMicId;
-  }
-}
-
-function handleMicrophoneChange() {
-  selectedMicrophoneId = microphoneSelect.value || null;
-  localStorage.setItem('selectedMicrophone', selectedMicrophoneId || '');
-
-  const selectedDevice = availableMicrophones.find(mic => mic.deviceId === selectedMicrophoneId);
-  const deviceName = selectedDevice ? selectedDevice.label : 'Default Microphone';
-
-  console.log(`Selected microphone: ${deviceName}`);
-  statusText.textContent = `Microphone changed to: ${deviceName}`;
-
-  if (isRecording) {
-    statusText.textContent = "Switching microphone... Please wait.";
-    stopRecording().then(() => {
-      setTimeout(() => {
-        toggleRecording();
-      }, 1000);
-    });
-  }
-}
-
-// Helpers
-function fmt1(x) {
-  const n = Number(x);
-  return Number.isFinite(n) ? n.toFixed(1) : x;
-}
-
-// Default WebSocket URL computation
-const host = window.location.hostname || "localhost";
-const port = window.location.port;
-const protocol = window.location.protocol === "https:" ? "wss" : "ws";
-const defaultWebSocketUrl = websocketUrl;
-
-// Populate default caption and input
-if (websocketDefaultSpan) websocketDefaultSpan.textContent = defaultWebSocketUrl;
-websocketInput.value = defaultWebSocketUrl;
-websocketUrl = defaultWebSocketUrl;
-
-// Optional chunk selector (guard for presence)
-if (chunkSelector) {
-  chunkSelector.addEventListener("change", () => {
-    chunkDuration = parseInt(chunkSelector.value);
-  });
-}
-
-// WebSocket input change handling
-websocketInput.addEventListener("change", () => {
-  const urlValue = websocketInput.value.trim();
-  if (!urlValue.startsWith("ws://") && !urlValue.startsWith("wss://")) {
-    statusText.textContent = "Invalid WebSocket URL (must start with ws:// or wss://)";
-    return;
-  }
-  websocketUrl = urlValue;
-  statusText.textContent = "WebSocket URL updated. Ready to connect.";
-});
-
-function setupWebSocket() {
-  return new Promise((resolve, reject) => {
-    try {
-      websocket = new WebSocket(websocketUrl);
-    } catch (error) {
-      statusText.textContent = "Invalid WebSocket URL. Please check and try again.";
-      reject(error);
-      return;
-    }
-
-    websocket.onopen = () => {
-      statusText.textContent = "Connected to server.";
-      resolve();
-    };
-
-    websocket.onclose = () => {
-      if (userClosing) {
-        if (waitingForStop) {
-          statusText.textContent = "Processing finalized or connection closed.";
-          if (lastReceivedData) {
-            renderLinesWithBuffer(
-              lastReceivedData.lines || [],
-              lastReceivedData.buffer_diarization || "",
-              lastReceivedData.buffer_transcription || "",
-              0,
-              0,
-              true
-            );
-          }
-        }
-      } else {
-        statusText.textContent = "Disconnected from the WebSocket server. (Check logs if model is loading.)";
-        if (isRecording) {
-          stopRecording();
-        }
-      }
-      isRecording = false;
-      waitingForStop = false;
-      userClosing = false;
-      lastReceivedData = null;
-      websocket = null;
-      updateUI();
-    };
-
-    websocket.onerror = () => {
-      statusText.textContent = "Error connecting to WebSocket.";
-      reject(new Error("Error connecting to WebSocket"));
-    };
-
-    websocket.onmessage = (event) => {
-      const data = JSON.parse(event.data);
-
-      if (data.type === "ready_to_stop") {
-        console.log("Ready to stop received, finalizing display and closing WebSocket.");
-        waitingForStop = false;
-
-        if (lastReceivedData) {
-          renderLinesWithBuffer(
-            lastReceivedData.lines || [],
-            lastReceivedData.buffer_diarization || "",
-            lastReceivedData.buffer_transcription || "",
-            0,
-            0,
-            true
-          );
-        }
-        statusText.textContent = "Finished processing audio! Ready to record again.";
-        recordButton.disabled = false;
-
-        if (websocket) {
-          websocket.close();
-        }
-        return;
-      }
-
-      lastReceivedData = data;
-
-      const {
-        lines = [],
-        buffer_transcription = "",
-        buffer_diarization = "",
-        remaining_time_transcription = 0,
-        remaining_time_diarization = 0,
-        status = "active_transcription",
-      } = data;
-
-      renderLinesWithBuffer(
-        lines,
-        buffer_diarization,
-        buffer_transcription,
-        remaining_time_diarization,
-        remaining_time_transcription,
-        false,
-        status
-      );
-    };
-  });
-}
-
-function renderLinesWithBuffer(
-  lines,
-  buffer_diarization,
-  buffer_transcription,
-  remaining_time_diarization,
-  remaining_time_transcription,
-  isFinalizing = false,
-  current_status = "active_transcription"
-) {
-  if (current_status === "no_audio_detected") {
-    linesTranscriptDiv.innerHTML =
-      "<p style='text-align: center; color: var(--muted); margin-top: 20px;'><em>No audio detected...</em></p>";
-    return;
-  }
-
-  const showLoading = !isFinalizing && (lines || []).some((it) => it.speaker == 0);
-  const showTransLag = !isFinalizing && remaining_time_transcription > 0;
-  const showDiaLag = !isFinalizing && !!buffer_diarization && remaining_time_diarization > 0;
-  const signature = JSON.stringify({
-    lines: (lines || []).map((it) => ({ speaker: it.speaker, text: it.text, start: it.start, end: it.end })),
-    buffer_transcription: buffer_transcription || "",
-    buffer_diarization: buffer_diarization || "",
-    status: current_status,
-    showLoading,
-    showTransLag,
-    showDiaLag,
-    isFinalizing: !!isFinalizing,
-  });
-  if (lastSignature === signature) {
-    const t = document.querySelector(".lag-transcription-value");
-    if (t) t.textContent = fmt1(remaining_time_transcription);
-    const d = document.querySelector(".lag-diarization-value");
-    if (d) d.textContent = fmt1(remaining_time_diarization);
-    const ld = document.querySelector(".loading-diarization-value");
-    if (ld) ld.textContent = fmt1(remaining_time_diarization);
-    return;
-  }
-  lastSignature = signature;
-
-  const linesHtml = (lines || [])
-    .map((item, idx) => {
-      let timeInfo = "";
-      if (item.start !== undefined && item.end !== undefined) {
-        timeInfo = ` ${item.start} - ${item.end}`;
-      }
-
-      let speakerLabel = "";
-      if (item.speaker === -2) {
-        speakerLabel = `<span class="silence">Silence<span id='timeInfo'>${timeInfo}</span></span>`;
-      } else if (item.speaker == 0 && !isFinalizing) {
-        speakerLabel = `<span class='loading'><span class="spinner"></span><span id='timeInfo'><span class="loading-diarization-value">${fmt1(
-          remaining_time_diarization
-        )}</span> second(s) of audio are undergoing diarization</span></span>`;
-      } else if (item.speaker !== 0) {
-        speakerLabel = `<span id="speaker">Speaker ${item.speaker}<span id='timeInfo'>${timeInfo}</span></span>`;
-      }
-
-      let currentLineText = item.text || "";
-
-      if (idx === lines.length - 1) {
-        if (!isFinalizing && item.speaker !== -2) {
-          if (remaining_time_transcription > 0) {
-            speakerLabel += `<span class="label_transcription"><span class="spinner"></span>Lag <span id='timeInfo'><span class="lag-transcription-value">${fmt1(
-              remaining_time_transcription
-            )}</span>s</span></span>`;
-          }
-          if (buffer_diarization && remaining_time_diarization > 0) {
-            speakerLabel += `<span class="label_diarization"><span class="spinner"></span>Lag<span id='timeInfo'><span class="lag-diarization-value">${fmt1(
-              remaining_time_diarization
-            )}</span>s</span></span>`;
-          }
-        }
-
-        if (buffer_diarization) {
-          if (isFinalizing) {
-            currentLineText +=
-              (currentLineText.length > 0 && buffer_diarization.trim().length > 0 ? " " : "") + buffer_diarization.trim();
-          } else {
-            currentLineText += `<span class="buffer_diarization">${buffer_diarization}</span>`;
-          }
-        }
-        if (buffer_transcription) {
-          if (isFinalizing) {
-            currentLineText +=
-              (currentLineText.length > 0 && buffer_transcription.trim().length > 0 ? " " : "") +
-              buffer_transcription.trim();
-          } else {
-            currentLineText += `<span class="buffer_transcription">${buffer_transcription}</span>`;
-          }
-        }
-      }
-
-      return currentLineText.trim().length > 0 || speakerLabel.length > 0
-        ? `<p>${speakerLabel}<br/><div class='textcontent'>${currentLineText}</div></p>`
-        : `<p>${speakerLabel}<br/></p>`;
-    })
-    .join("");
-
-  linesTranscriptDiv.innerHTML = linesHtml;
-  window.scrollTo({ top: document.body.scrollHeight, behavior: "smooth" });
-}
-
-function updateTimer() {
-  if (!startTime) return;
-
-  const elapsed = Math.floor((Date.now() - startTime) / 1000);
-  const minutes = Math.floor(elapsed / 60).toString().padStart(2, "0");
-  const seconds = (elapsed % 60).toString().padStart(2, "0");
-  timerElement.textContent = `${minutes}:${seconds}`;
-}
-
-function drawWaveform() {
-  if (!analyser) return;
-
-  const bufferLength = analyser.frequencyBinCount;
-  const dataArray = new Uint8Array(bufferLength);
-  analyser.getByteTimeDomainData(dataArray);
-
-  waveCtx.clearRect(
-    0,
-    0,
-    waveCanvas.width / (window.devicePixelRatio || 1),
-    waveCanvas.height / (window.devicePixelRatio || 1)
-  );
-  waveCtx.lineWidth = 1;
-  waveCtx.strokeStyle = waveStroke;
-  waveCtx.beginPath();
-
-  const sliceWidth = (waveCanvas.width / (window.devicePixelRatio || 1)) / bufferLength;
-  let x = 0;
-
-  for (let i = 0; i < bufferLength; i++) {
-    const v = dataArray[i] / 128.0;
-    const y = (v * (waveCanvas.height / (window.devicePixelRatio || 1))) / 2;
-
-    if (i === 0) {
-      waveCtx.moveTo(x, y);
-    } else {
-      waveCtx.lineTo(x, y);
-    }
-
-    x += sliceWidth;
-  }
-
-  waveCtx.lineTo(
-    waveCanvas.width / (window.devicePixelRatio || 1),
-    (waveCanvas.height / (window.devicePixelRatio || 1)) / 2
-  );
-  waveCtx.stroke();
-
-  animationFrame = requestAnimationFrame(drawWaveform);
-}
-
-async function startRecording() {
-  try {
-    try {
-      wakeLock = await navigator.wakeLock.request("screen");
-    } catch (err) {
-      console.log("Error acquiring wake lock.");
-    }
-
-    let stream;
-    try {
-      // Try tab capture first
-      stream = await new Promise((resolve, reject) => {
-        chrome.tabCapture.capture({audio: true}, (s) => {
-          if (s) {
-            resolve(s);
-          } else {
-            reject(new Error('Tab capture failed or not available'));
-          }
-        });
-      });
-      statusText.textContent = "Using tab audio capture.";
-    } catch (tabError) {
-      console.log('Tab capture not available, falling back to microphone', tabError);
-      // Fallback to microphone
-      const audioConstraints = selectedMicrophoneId
-        ? { audio: { deviceId: { exact: selectedMicrophoneId } } }
-        : { audio: true };
-      stream = await navigator.mediaDevices.getUserMedia(audioConstraints);
-      statusText.textContent = "Using microphone audio.";
-    }
-
-    audioContext = new (window.AudioContext || window.webkitAudioContext)();
-    analyser = audioContext.createAnalyser();
-    analyser.fftSize = 256;
-    microphone = audioContext.createMediaStreamSource(stream);
-    microphone.connect(analyser);
-
-    recorder = new MediaRecorder(stream, { mimeType: "audio/webm" });
-    recorder.ondataavailable = (e) => {
-      if (websocket && websocket.readyState === WebSocket.OPEN) {
-        websocket.send(e.data);
-      }
-    };
-    recorder.start(chunkDuration);
-
-    startTime = Date.now();
-    timerInterval = setInterval(updateTimer, 1000);
-    drawWaveform();
-
-    isRecording = true;
-    updateUI();
-  } catch (err) {
-    if (window.location.hostname === "0.0.0.0") {
-      statusText.textContent =
-        "Error accessing audio input. Browsers may block audio access on 0.0.0.0. Try using localhost:8000 instead.";
-    } else {
-      statusText.textContent = "Error accessing audio input. Please check permissions.";
-    }
-    console.error(err);
-  }
-}
-
-async function stopRecording() {
-  if (wakeLock) {
-    try {
-      await wakeLock.release();
-    } catch (e) {
-      // ignore
-    }
-    wakeLock = null;
-  }
-
-  userClosing = true;
-  waitingForStop = true;
-
-  if (websocket && websocket.readyState === WebSocket.OPEN) {
-    const emptyBlob = new Blob([], { type: "audio/webm" });
-    websocket.send(emptyBlob);
-    statusText.textContent = "Recording stopped. Processing final audio...";
-  }
-
-  if (recorder) {
-    recorder.stop();
-    recorder = null;
-  }
-
-  if (microphone) {
-    microphone.disconnect();
-    microphone = null;
-  }
-
-  if (analyser) {
-    analyser = null;
-  }
-
-  if (audioContext && audioContext.state !== "closed") {
-    try {
-      await audioContext.close();
-    } catch (e) {
-      console.warn("Could not close audio context:", e);
-    }
-    audioContext = null;
-  }
-
-  if (animationFrame) {
-    cancelAnimationFrame(animationFrame);
-    animationFrame = null;
-  }
-
-  if (timerInterval) {
-    clearInterval(timerInterval);
-    timerInterval = null;
-  }
-  timerElement.textContent = "00:00";
-  startTime = null;
-
-  isRecording = false;
-  updateUI();
-}
-
-async function toggleRecording() {
-  if (!isRecording) {
-    if (waitingForStop) {
-      console.log("Waiting for stop, early return");
-      return;
-    }
-    console.log("Connecting to WebSocket");
-    try {
-      if (websocket && websocket.readyState === WebSocket.OPEN) {
-        await startRecording();
-      } else {
-        await setupWebSocket();
-        await startRecording();
-      }
-    } catch (err) {
-      statusText.textContent = "Could not connect to WebSocket or access mic. Aborted.";
-      console.error(err);
-    }
-  } else {
-    console.log("Stopping recording");
-    stopRecording();
-  }
-}
-
-function updateUI() {
-  recordButton.classList.toggle("recording", isRecording);
-  recordButton.disabled = waitingForStop;
-
-  if (waitingForStop) {
-    if (statusText.textContent !== "Recording stopped. Processing final audio...") {
-      statusText.textContent = "Please wait for processing to complete...";
-    }
-  } else if (isRecording) {
-    statusText.textContent = "Recording...";
-  } else {
-    if (
-      statusText.textContent !== "Finished processing audio! Ready to record again." &&
-      statusText.textContent !== "Processing finalized or connection closed."
-    ) {
-      statusText.textContent = "Click to start transcription";
-    }
-  }
-  if (!waitingForStop) {
-    recordButton.disabled = false;
-  }
-}
-
-recordButton.addEventListener("click", toggleRecording);
-
-if (microphoneSelect) {
-  microphoneSelect.addEventListener("change", handleMicrophoneChange);
-}
-
-// Settings toggle functionality
-settingsToggle.addEventListener("click", () => {
-  settingsDiv.classList.toggle("visible");
-  settingsToggle.classList.toggle("active");
-});
-
-document.addEventListener('DOMContentLoaded', async () => {
-  try {
-    await enumerateMicrophones();
-  } catch (error) {
-    console.log("Could not enumerate microphones on load:", error);
-  }
-});
-navigator.mediaDevices.addEventListener('devicechange', async () => {
-  console.log('Device change detected, re-enumerating microphones');
-  try {
-    await enumerateMicrophones();
-  } catch (error) {
-    console.log("Error re-enumerating microphones:", error);
-  }
-});
-
-
-async function run() {
-  const micPermission = await navigator.permissions.query({
-    name: "microphone",
-  });
-
-  document.getElementById(
-    "audioPermission"
-  ).innerText = `MICROPHONE: ${micPermission.state}`;
-
-  if (micPermission.state !== "granted") {
-    chrome.tabs.create({ url: "welcome.html" });
-  }
-
-  const intervalId = setInterval(async () => {
-    const micPermission = await navigator.permissions.query({
-      name: "microphone",
-    });
-    if (micPermission.state === "granted") {
-      document.getElementById(
-        "audioPermission"
-      ).innerText = `MICROPHONE: ${micPermission.state}`;
-      clearInterval(intervalId);
-    }
-  }, 100);
-}
-
-void run();

chrome-extension/manifest.json

@@ -3,9 +3,6 @@
     "name": "WhisperLiveKit Tab Capture",
     "version": "1.0",
     "description": "Capture and transcribe audio from browser tabs using WhisperLiveKit.",
-    "background": {
-        "service_worker": "background.js"
-    },
     "icons": {
         "16": "icons/icon16.png",
         "32": "icons/icon32.png",
@@ -14,7 +11,7 @@
     },
     "action": {
         "default_title": "WhisperLiveKit Tab Capture",
-        "default_popup": "popup.html"
+        "default_popup": "live_transcription.html"
     },
     "permissions": [
         "scripting",
@@ -22,16 +19,5 @@
         "offscreen",
         "activeTab",
         "storage"
-    ],
-    "web_accessible_resources": [
-        {
-            "resources": [
-                "requestPermissions.html",
-                "requestPermissions.js"
-            ],
-            "matches": [
-                "<all_urls>"
-            ]
-        }
     ]
 }

chrome-extension/popup.html

@@ -1,78 +0,0 @@
-<!DOCTYPE html>
-<html lang="en">
-
-<head>
-    <meta charset="UTF-8" />
-    <meta name="viewport" content="width=device-width, initial-scale=1.0" />
-    <title>WhisperLiveKit</title>
-    <link rel="stylesheet" href="/web/live_transcription.css" />
-</head>
-
-<body>
-    <div class="settings-container">
-        <button id="recordButton">
-            <div class="shape-container">
-                <div class="shape"></div>
-            </div>
-            <div class="recording-info">
-                <div class="wave-container">
-                    <canvas id="waveCanvas"></canvas>
-                </div>
-                <div class="timer">00:00</div>
-            </div>
-        </button>
-
-        <button id="settingsToggle" class="settings-toggle" title="Show/hide settings">
-            <img src="/web/src/settings.svg" alt="Settings" />
-        </button>
-
-        <div class="settings">
-            <div class="field">
-                <label for="websocketInput">Websocket URL</label>
-                <input id="websocketInput" type="text" placeholder="ws://host:port/asr" />
-            </div>
-
-            <div class="field">
-                <label id="microphoneSelectLabel" for="microphoneSelect">Select Microphone</label>
-                <select id="microphoneSelect">
-                    <option value="">Default Microphone</option>
-                </select>
-                        <div id="audioPermission"></div>
-
-            </div>
-
-            <div class="theme-selector-container">
-                <div class="segmented" role="radiogroup" aria-label="Theme selector">
-                    <input type="radio" id="theme-system" name="theme" value="system" />
-                    <label for="theme-system" title="System">
-                        <img src="/web/src/system_mode.svg" alt="" />
-                        <!-- <span>System</span> -->
-                    </label>
-
-                    <input type="radio" id="theme-light" name="theme" value="light" />
-                    <label for="theme-light" title="Light">
-                        <img src="/web/src/light_mode.svg" alt="" />
-                        <!-- <span>Light</span> -->
-                    </label>
-
-                    <input type="radio" id="theme-dark" name="theme" value="dark" />
-                    <label for="theme-dark" title="Dark">
-                        <img src="/web/src/dark_mode.svg" alt="" />
-                        <!-- <span>Dark</span> -->
-                    </label>
-                </div>
-            </div>
-
-        </div>
-    </div>
-
-
-
-    <p id="status"></p>
-
-    <div id="linesTranscript"></div>
-
-    <script src="live_transcription.js"></script>
-</body>
-
-</html>

chrome-extension/web/live_transcription.css

@@ -1,539 +0,0 @@
-:root {
-  --bg: #ffffff;
-  --text: #111111;
-  --muted: #666666;
-  --border: #e5e5e5;
-  --chip-bg: rgba(0, 0, 0, 0.04);
-  --chip-text: #000000;
-  --spinner-border: #8d8d8d5c;
-  --spinner-top: #b0b0b0;
-  --silence-bg: #f3f3f3;
-  --loading-bg: rgba(255, 77, 77, 0.06);
-  --button-bg: #ffffff;
-  --button-border: #e9e9e9;
-  --wave-stroke: #000000;
-  --label-dia-text: #868686;
-  --label-trans-text: #111111;
-}
-
-@media (prefers-color-scheme: dark) {
-  :root:not([data-theme="light"]) {
-    --bg: #0b0b0b;
-    --text: #e6e6e6;
-    --muted: #9aa0a6;
-    --border: #333333;
-    --chip-bg: rgba(255, 255, 255, 0.08);
-    --chip-text: #e6e6e6;
-    --spinner-border: #555555;
-    --spinner-top: #dddddd;
-    --silence-bg: #1a1a1a;
-    --loading-bg: rgba(255, 77, 77, 0.12);
-    --button-bg: #111111;
-    --button-border: #333333;
-    --wave-stroke: #e6e6e6;
-    --label-dia-text: #b3b3b3;
-    --label-trans-text: #ffffff;
-  }
-}
-
-:root[data-theme="dark"] {
-  --bg: #0b0b0b;
-  --text: #e6e6e6;
-  --muted: #9aa0a6;
-  --border: #333333;
-  --chip-bg: rgba(255, 255, 255, 0.08);
-  --chip-text: #e6e6e6;
-  --spinner-border: #555555;
-  --spinner-top: #dddddd;
-  --silence-bg: #1a1a1a;
-  --loading-bg: rgba(255, 77, 77, 0.12);
-  --button-bg: #111111;
-  --button-border: #333333;
-  --wave-stroke: #e6e6e6;
-  --label-dia-text: #b3b3b3;
-  --label-trans-text: #ffffff;
-}
-
-:root[data-theme="light"] {
-  --bg: #ffffff;
-  --text: #111111;
-  --muted: #666666;
-  --border: #e5e5e5;
-  --chip-bg: rgba(0, 0, 0, 0.04);
-  --chip-text: #000000;
-  --spinner-border: #8d8d8d5c;
-  --spinner-top: #b0b0b0;
-  --silence-bg: #f3f3f3;
-  --loading-bg: rgba(255, 77, 77, 0.06);
-  --button-bg: #ffffff;
-  --button-border: #e9e9e9;
-  --wave-stroke: #000000;
-  --label-dia-text: #868686;
-  --label-trans-text: #111111;
-}
-
-body {
-  font-family: ui-sans-serif, system-ui, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji';
-  margin: 20px;
-  text-align: center;
-  background-color: var(--bg);
-  color: var(--text);
-}
-
-.settings-toggle {
-  margin-top: 4px;
-  width: 40px;
-  height: 40px;
-  border: none;
-  border-radius: 50%;
-  background-color: var(--button-bg);
-  cursor: pointer;
-  transition: all 0.3s ease;
-  /* border: 1px solid var(--button-border); */
-  display: flex;
-  align-items: center;
-  justify-content: center;
-  position: relative;
-}
-
-.settings-toggle:hover {
-  background-color: var(--chip-bg);
-}
-
-.settings-toggle img {
-  width: 24px;
-  height: 24px;
-  opacity: 0.7;
-  transition: opacity 0.2s ease, transform 0.3s ease;
-}
-
-.settings-toggle:hover img {
-  opacity: 1;
-}
-
-.settings-toggle.active img {
-  transform: rotate(80deg);
-}
-
-/* Record button */
-#recordButton {
-  width: 50px;
-  height: 50px;
-  border: none;
-  border-radius: 50%;
-  background-color: var(--button-bg);
-  cursor: pointer;
-  transition: all 0.3s ease;
-  border: 1px solid var(--button-border);
-  display: flex;
-  align-items: center;
-  justify-content: center;
-  position: relative;
-}
-
-#recordButton.recording {
-  width: 180px;
-  border-radius: 40px;
-  justify-content: flex-start;
-  padding-left: 20px;
-}
-
-#recordButton:active {
-  transform: scale(0.95);
-}
-
-.shape-container {
-  width: 25px;
-  height: 25px;
-  display: flex;
-  align-items: center;
-  justify-content: center;
-  flex-shrink: 0;
-}
-
-.shape {
-  width: 25px;
-  height: 25px;
-  background-color: rgb(209, 61, 53);
-  border-radius: 50%;
-  transition: all 0.3s ease;
-}
-
-#recordButton:disabled .shape {
-  background-color: #6e6d6d;
-}
-
-#recordButton.recording .shape {
-  border-radius: 5px;
-  width: 25px;
-  height: 25px;
-}
-
-/* Recording elements */
-.recording-info {
-  display: none;
-  align-items: center;
-  margin-left: 15px;
-  flex-grow: 1;
-}
-
-#recordButton.recording .recording-info {
-  display: flex;
-}
-
-.wave-container {
-  width: 60px;
-  height: 30px;
-  position: relative;
-  display: flex;
-  align-items: center;
-  justify-content: center;
-}
-
-#waveCanvas {
-  width: 100%;
-  height: 100%;
-}
-
-.timer {
-  font-size: 14px;
-  font-weight: 500;
-  color: var(--text);
-  margin-left: 10px;
-}
-
-#status {
-  margin-top: 20px;
-  font-size: 16px;
-  color: var(--text);
-}
-
-/* Settings */
-.settings-container {
-  display: flex;
-  justify-content: center;
-  align-items: flex-start;
-  gap: 15px;
-  margin-top: 20px;
-  flex-wrap: wrap;
-}
-
-.settings {
-  display: none;
-  flex-wrap: wrap;
-  align-items: flex-start;
-  gap: 12px;
-  transition: opacity 0.3s ease;
-}
-
-.settings.visible {
-  display: flex;
-}
-
-.field {
-  display: flex;
-  flex-direction: column;
-  align-items: flex-start;
-  gap: 3px;
-}
-
-#chunkSelector,
-#websocketInput,
-#themeSelector,
-#microphoneSelect {
-  font-size: 16px;
-  padding: 5px 8px;
-  border-radius: 8px;
-  border: 1px solid var(--border);
-  background-color: var(--button-bg);
-  color: var(--text);
-  max-height: 30px;
-}
-
-#microphoneSelect {
-  width: 100%;
-  max-width: 190px;
-  min-width: 120px;
-}
-
-#chunkSelector:focus,
-#websocketInput:focus,
-#themeSelector:focus,
-#microphoneSelect:focus {
-  outline: none;
-  border-color: #007bff;
-  box-shadow: 0 0 0 3px rgba(0, 123, 255, 0.15);
-}
-
-label {
-  font-size: 13px;
-  color: var(--muted);
-}
-
-.ws-default {
-  font-size: 12px;
-  color: var(--muted);
-}
-
-/* Segmented pill control for Theme */
-.segmented {
-  display: inline-flex;
-  align-items: stretch;
-  border: 1px solid var(--button-border);
-  background-color: var(--button-bg);
-  border-radius: 999px;
-  overflow: hidden;
-}
-
-.segmented input[type="radio"] {
-  position: absolute;
-  opacity: 0;
-  pointer-events: none;
-}
-
-.theme-selector-container {
-  display: flex;
-  align-items: center;
-  margin-top: 17px;
-}
-
-.segmented label {
-  display: inline-flex;
-  align-items: center;
-  gap: 6px;
-  padding: 6px 12px;
-  font-size: 14px;
-  color: var(--muted);
-  cursor: pointer;
-  user-select: none;
-  transition: background-color 0.2s ease, color 0.2s ease;
-}
-
-.segmented label span {
-  display: none;
-}
-
-.segmented label:hover span {
-  display: inline;
-}
-
-.segmented label:hover {
-  background-color: var(--chip-bg);
-}
-
-.segmented img {
-  width: 16px;
-  height: 16px;
-}
-
-.segmented input[type="radio"]:checked + label {
-  background-color: var(--chip-bg);
-  color: var(--text);
-}
-
-.segmented input[type="radio"]:focus-visible + label,
-.segmented input[type="radio"]:focus + label {
-  outline: 2px solid #007bff;
-  outline-offset: 2px;
-  border-radius: 999px;
-}
-
-/* Transcript area */
-#linesTranscript {
-  margin: 20px auto;
-  max-width: 700px;
-  text-align: left;
-  font-size: 16px;
-}
-
-#linesTranscript p {
-  margin: 0px 0;
-}
-
-#linesTranscript strong {
-  color: var(--text);
-}
-
-#speaker {
-  border: 1px solid var(--border);
-  border-radius: 100px;
-  padding: 2px 10px;
-  font-size: 14px;
-  margin-bottom: 0px;
-}
-
-.label_diarization {
-  background-color: var(--chip-bg);
-  border-radius: 8px 8px 8px 8px;
-  padding: 2px 10px;
-  margin-left: 10px;
-  display: inline-block;
-  white-space: nowrap;
-  font-size: 14px;
-  margin-bottom: 0px;
-  color: var(--label-dia-text);
-}
-
-.label_transcription {
-  background-color: var(--chip-bg);
-  border-radius: 8px 8px 8px 8px;
-  padding: 2px 10px;
-  display: inline-block;
-  white-space: nowrap;
-  margin-left: 10px;
-  font-size: 14px;
-  margin-bottom: 0px;
-  color: var(--label-trans-text);
-}
-
-#timeInfo {
-  color: var(--muted);
-  margin-left: 10px;
-}
-
-.textcontent {
-  font-size: 16px;
-  padding-left: 10px;
-  margin-bottom: 10px;
-  margin-top: 1px;
-  padding-top: 5px;
-  border-radius: 0px 0px 0px 10px;
-}
-
-.buffer_diarization {
-  color: var(--label-dia-text);
-  margin-left: 4px;
-}
-
-.buffer_transcription {
-  color: #7474748c;
-  margin-left: 4px;
-}
-
-.spinner {
-  display: inline-block;
-  width: 8px;
-  height: 8px;
-  border: 2px solid var(--spinner-border);
-  border-top: 2px solid var(--spinner-top);
-  border-radius: 50%;
-  animation: spin 0.7s linear infinite;
-  vertical-align: middle;
-  margin-bottom: 2px;
-  margin-right: 5px;
-}
-
-@keyframes spin {
-  to {
-    transform: rotate(360deg);
-  }
-}
-
-.silence {
-  color: var(--muted);
-  background-color: var(--silence-bg);
-  font-size: 13px;
-  border-radius: 30px;
-  padding: 2px 10px;
-}
-
-.loading {
-  color: var(--muted);
-  background-color: var(--loading-bg);
-  border-radius: 8px 8px 8px 0px;
-  padding: 2px 10px;
-  font-size: 14px;
-  margin-bottom: 0px;
-}
-
-/* for smaller screens */
-/* @media (max-width: 450px) {
-  .settings-container {
-    flex-direction: column;
-    gap: 10px;
-    align-items: center;
-  }
-
-  .settings {
-    justify-content: center;
-    gap: 8px;
-    width: 100%;
-  }
-
-  .field {
-    align-items: center;
-    width: 100%;
-  }
-
-  #websocketInput,
-  #microphoneSelect {
-    min-width: 200px;
-    max-width: 100%;
-  }
-
-  .theme-selector-container {
-    margin-top: 10px;
-  }
-} */
-
-/* @media (max-width: 768px) and (min-width: 451px) {
-  .settings-container {
-    gap: 10px;
-  }
-
-  .settings {
-    gap: 8px;
-  }
-
-  #websocketInput,
-  #microphoneSelect {
-    min-width: 150px;
-    max-width: 300px;
-  }
-} */
-
-/* @media (max-width: 480px) {
-  body {
-    margin: 10px;
-  }
-
-  .settings-toggle {
-    width: 35px;
-    height: 35px;
-  }
-
-  .settings-toggle img {
-    width: 20px;
-    height: 20px;
-  }
-
-  .settings {
-    flex-direction: column;
-    align-items: center;
-    gap: 6px;
-  }
-
-  #websocketInput,
-  #microphoneSelect {
-    max-width: 400px;
-  }
-
-  .segmented label {
-    padding: 4px 8px;
-    font-size: 12px;
-  }
-
-  .segmented img {
-    width: 14px;
-    height: 14px;
-  }
-} */
-
-
-html
-{
-    width: 400px;  /* max: 800px */
-    height: 600px; /* max: 600px */
-    border-radius: 10px;
-
-}

chrome-extension/web/src/dark_mode.svg

@@ -1 +0,0 @@
-<svg xmlns="http://www.w3.org/2000/svg" height="24px" viewBox="0 -960 960 960" width="24px" fill="#5f6368"><path d="M480-120q-151 0-255.5-104.5T120-480q0-138 90-239.5T440-838q13-2 23 3.5t16 14.5q6 9 6.5 21t-7.5 23q-17 26-25.5 55t-8.5 61q0 90 63 153t153 63q31 0 61.5-9t54.5-25q11-7 22.5-6.5T819-479q10 5 15.5 15t3.5 24q-14 138-117.5 229T480-120Zm0-80q88 0 158-48.5T740-375q-20 5-40 8t-40 3q-123 0-209.5-86.5T364-660q0-20 3-40t8-40q-78 32-126.5 102T200-480q0 116 82 198t198 82Zm-10-270Z"/></svg>

chrome-extension/web/src/light_mode.svg

@@ -1 +0,0 @@
-<svg xmlns="http://www.w3.org/2000/svg" height="24px" viewBox="0 -960 960 960" width="24px" fill="#5f6368"><path d="M480-360q50 0 85-35t35-85q0-50-35-85t-85-35q-50 0-85 35t-35 85q0 50 35 85t85 35Zm0 80q-83 0-141.5-58.5T280-480q0-83 58.5-141.5T480-680q83 0 141.5 58.5T680-480q0 83-58.5 141.5T480-280ZM80-440q-17 0-28.5-11.5T40-480q0-17 11.5-28.5T80-520h80q17 0 28.5 11.5T200-480q0 17-11.5 28.5T160-440H80Zm720 0q-17 0-28.5-11.5T760-480q0-17 11.5-28.5T800-520h80q17 0 28.5 11.5T920-480q0 17-11.5 28.5T880-440h-80ZM480-760q-17 0-28.5-11.5T440-800v-80q0-17 11.5-28.5T480-920q17 0 28.5 11.5T520-880v80q0 17-11.5 28.5T480-760Zm0 720q-17 0-28.5-11.5T440-80v-80q0-17 11.5-28.5T480-200q17 0 28.5 11.5T520-160v80q0 17-11.5 28.5T480-40ZM226-678l-43-42q-12-11-11.5-28t11.5-29q12-12 29-12t28 12l42 43q11 12 11 28t-11 28q-11 12-27.5 11.5T226-678Zm494 495-42-43q-11-12-11-28.5t11-27.5q11-12 27.5-11.5T734-282l43 42q12 11 11.5 28T777-183q-12 12-29 12t-28-12Zm-42-495q-12-11-11.5-27.5T678-734l42-43q11-12 28-11.5t29 11.5q12 12 12 29t-12 28l-43 42q-12 11-28 11t-28-11ZM183-183q-12-12-12-29t12-28l43-42q12-11 28.5-11t27.5 11q12 11 11.5 27.5T282-226l-42 43q-11 12-28 11.5T183-183Zm297-297Z"/></svg>

chrome-extension/web/src/settings.svg

@@ -1 +0,0 @@
-<svg xmlns="http://www.w3.org/2000/svg" height="24px" viewBox="0 -960 960 960" width="24px" fill="#5f6368"><path d="M433-80q-27 0-46.5-18T363-142l-9-66q-13-5-24.5-12T307-235l-62 26q-25 11-50 2t-39-32l-47-82q-14-23-8-49t27-43l53-40q-1-7-1-13.5v-27q0-6.5 1-13.5l-53-40q-21-17-27-43t8-49l47-82q14-23 39-32t50 2l62 26q11-8 23-15t24-12l9-66q4-26 23.5-44t46.5-18h94q27 0 46.5 18t23.5 44l9 66q13 5 24.5 12t22.5 15l62-26q25-11 50-2t39 32l47 82q14 23 8 49t-27 43l-53 40q1 7 1 13.5v27q0 6.5-2 13.5l53 40q21 17 27 43t-8 49l-48 82q-14 23-39 32t-50-2l-60-26q-11 8-23 15t-24 12l-9 66q-4 26-23.5 44T527-80h-94Zm7-80h79l14-106q31-8 57.5-23.5T639-327l99 41 39-68-86-65q5-14 7-29.5t2-31.5q0-16-2-31.5t-7-29.5l86-65-39-68-99 42q-22-23-48.5-38.5T533-694l-13-106h-79l-14 106q-31 8-57.5 23.5T321-633l-99-41-39 68 86 64q-5 15-7 30t-2 32q0 16 2 31t7 30l-86 65 39 68 99-42q22 23 48.5 38.5T427-266l13 106Zm42-180q58 0 99-41t41-99q0-58-41-99t-99-41q-59 0-99.5 41T342-480q0 58 40.5 99t99.5 41Zm-2-140Z"/></svg>

chrome-extension/web/src/system_mode.svg

@@ -1 +0,0 @@
-<svg xmlns="http://www.w3.org/2000/svg" height="24px" viewBox="0 -960 960 960" width="24px" fill="#5f6368"><path d="M396-396q-32-32-58.5-67T289-537q-5 14-6.5 28.5T281-480q0 83 58 141t141 58q14 0 28.5-2t28.5-6q-39-22-74-48.5T396-396Zm85 196q-56 0-107-21t-91-61q-40-40-61-91t-21-107q0-51 17-97.5t50-84.5q13-14 32-9.5t27 24.5q21 55 52.5 104t73.5 91q42 42 91 73.5T648-326q20 8 24.5 27t-9.5 32q-38 33-84.5 50T481-200Zm223-192q-16-5-23-20.5t-4-32.5q9-48-6-94.5T621-621q-35-35-80.5-49.5T448-677q-17 3-32-4t-21-23q-6-16 1.5-31t23.5-19q69-15 138 4.5T679-678q51 51 71 120t5 138q-4 17-19 25t-32 3ZM480-840q-17 0-28.5-11.5T440-880v-40q0-17 11.5-28.5T480-960q17 0 28.5 11.5T520-920v40q0 17-11.5 28.5T480-840Zm0 840q-17 0-28.5-11.5T440-40v-40q0-17 11.5-28.5T480-120q17 0 28.5 11.5T520-80v40q0 17-11.5 28.5T480 0Zm255-734q-12-12-12-28.5t12-28.5l28-28q11-11 27.5-11t28.5 11q12 12 12 28.5T819-762l-28 28q-12 12-28 12t-28-12ZM141-141q-12-12-12-28.5t12-28.5l28-28q12-12 28-12t28 12q12 12 12 28.5T225-169l-28 28q-11 11-27.5 11T141-141Zm739-299q-17 0-28.5-11.5T840-480q0-17 11.5-28.5T880-520h40q17 0 28.5 11.5T960-480q0 17-11.5 28.5T920-440h-40Zm-840 0q-17 0-28.5-11.5T0-480q0-17 11.5-28.5T40-520h40q17 0 28.5 11.5T120-480q0 17-11.5 28.5T80-440H40Zm779 299q-12 12-28.5 12T762-141l-28-28q-12-12-12-28t12-28q12-12 28.5-12t28.5 12l28 28q11 11 11 27.5T819-141ZM226-735q-12 12-28.5 12T169-735l-28-28q-11-11-11-27.5t11-28.5q12-12 28.5-12t28.5 12l28 28q12 12 12 28t-12 28Zm170 339Z"/></svg>

chrome-extension/welcome.html

@@ -1,12 +0,0 @@
-<!DOCTYPE html>
-<html>
-  <head>
-    <title>Welcome</title>
-    <script src="welcome.js"></script>
-  </head>
-  <body>
-    This page exists to workaround an issue with Chrome that blocks permission
-    requests from chrome extensions
-    <!-- <button id="requestMicrophone">Request Microphone</button> -->
-  </body>
-</html>

docs/API.md

@@ -0,0 +1,264 @@
+# WhisperLiveKit WebSocket API Documentation
+
+> !! **Note**: The new API structure described in this document is currently under deployment. 
+This documentation is intended for devs who want to build custom frontends.
+
+WLK provides real-time speech transcription, speaker diarization, and translation through a WebSocket API. The server sends incremental updates as audio is processed, allowing clients to display live transcription results with minimal latency.
+
+---
+
+## Legacy API (Current)
+
+### Message Structure
+
+The current API sends complete state snapshots on each update (several time per second)
+
+```typescript
+{
+  "type": str,
+  "status": str,
+  "lines": [
+    {
+      "speaker": int,
+      "text": str,
+      "start": float,
+      "end": float,
+      "translation": str | null,
+      "detected_language": str
+    }
+  ],
+  "buffer_transcription": str,
+  "buffer_diarization": str,
+  "remaining_time_transcription": float,
+  "remaining_time_diarization": float
+}
+```
+
+---
+
+## New API (Under Development)
+
+### Philosophy
+
+Principles:
+
+- **Incremental Updates**: Only updates and new segments are sent
+- **Ephemeral Buffers**: Temporary, unvalidated data displayed in real-time but overwritten on next update, at speaker level
+
+
+## Message Format
+
+
+```typescript
+{
+  "type": "transcript_update",
+  "status": "active_transcription" | "no_audio_detected",
+  "segments": [
+    {
+      "id": number,
+      "speaker": number,
+      "text": string,
+      "start_speaker": float,
+      "start": float,
+      "end": float,
+      "language": string | null,
+      "translation": string,
+      "words": [
+        {
+          "text": string,
+          "start": float,
+          "end": float,
+          "validated": {
+            "text": boolean,
+            "speaker": boolean,
+          }
+        }
+      ],
+      "buffer": {
+        "transcription": string,
+        "diarization": string,
+        "translation": string
+      }
+    }
+  ],
+  "metadata": {
+    "remaining_time_transcription": float,
+    "remaining_time_diarization": float
+  }
+}
+```
+
+### Other Message Types
+
+#### Config Message (sent on connection)
+```json
+{
+  "type": "config",
+  "useAudioWorklet": true / false
+}
+```
+
+#### Ready to Stop Message (sent after processing complete)
+```json
+{
+  "type": "ready_to_stop"
+}
+```
+
+---
+
+## Field Descriptions
+
+### Segment Fields
+
+| Field | Type | Description |
+|-------|------|-------------|
+| `id` | `number` | Unique identifier for this segment. Used by clients to update specific segments efficiently. |
+| `speaker` | `number` | Speaker ID (1, 2, 3...). Special value `-2` indicates silence. |
+| `text` | `string` | Validated transcription text for this update. Should be **appended** to the segment's text on the client side.  |
+| `start_speaker` | `float` | Timestamp (seconds) when this speaker segment began. |
+| `start` | `float` | Timestamp (seconds) of the first word in this update. |
+| `end` | `float` | Timestamp (seconds) of the last word in this update. |
+| `language` | `string \| null` | ISO language code (e.g., "en", "fr"). `null` until language is detected. |
+| `translation` | `string` | Validated translation text for this update. Should be **appended** to the segment's translation on the client side. |
+| `words` | `Array` | Array of word-level objects with timing and validation information. |
+| `buffer` | `Object` | Per-segment temporary buffers, see below |
+
+### Word Object
+
+| Field | Type | Description |
+|-------|------|-------------|
+| `text` | `string` | The word text. |
+| `start` | `number` | Start timestamp (seconds) of this word. |
+| `end` | `number` | End timestamp (seconds) of this word. |
+| `validated.text` | `boolean` | Whether the transcription text has been validated. if false, word is also in buffer: transcription |
+| `validated.speaker` | `boolean` | Whether the speaker assignment has been validated. if false, word is also in buffer: diarization |
+| `validated.language` | `boolean` | Whether the language detection has been validated. if false, word is also in buffer: translation |
+
+### Buffer Object (Per-Segment)
+
+Buffers are **ephemeral**. They should be displayed to the user but not stored permanently in the frontend. Each update may contain a completely different buffer value, and previous buffer is likely to be in the next validated text.
+
+| Field | Type | Description |
+|-------|------|-------------|
+| `transcription` | `string` | Pending transcription text. Displayed immediately but **overwritten** on next update. |
+| `diarization` | `string` | Pending diarization text (text waiting for speaker assignment). Displayed immediately but **overwritten** on next update. |
+| `translation` | `string` | Pending translation text. Displayed immediately but **overwritten** on next update. |
+
+
+### Metadata Fields
+
+| Field | Type | Description |
+|-------|------|-------------|
+| `remaining_time_transcription` | `float` | Seconds of audio waiting for transcription processing. |
+| `remaining_time_diarization` | `float` | Seconds of audio waiting for speaker diarization. |
+
+### Status Values
+
+| Status | Description |
+|--------|-------------|
+| `active_transcription` | Normal operation, transcription is active. |
+| `no_audio_detected` | No audio has been detected yet. |
+
+---
+
+## Update Behavior
+
+### Incremental Updates
+
+The API sends **only changed or new segments**. Clients should:
+
+1. Maintain a local map of segments by ID
+2. When receiving an update, merge/update segments by ID
+3. Render only the changed segments
+
+### Language Detection
+
+When language is detected for a segment:
+
+```jsonc
+// Update 1: No language yet
+{
+  "segments": [
+    {"id": 1, "speaker": 1, "text": "May see", "language": null}
+  ]
+}
+
+// Update 2: Same segment ID, language now detected
+{
+  "segments": [
+    {"id": 1, "speaker": 1, "text": "Merci", "language": "fr"}
+  ]
+}
+```
+
+**Client behavior**: **Replace** the existing segment with the same ID.
+
+### Buffer Behavior
+
+Buffers are **per-segment** to handle multi-speaker scenarios correctly.
+
+#### Example: Translation with diarization and translation
+
+```jsonc
+// Update 1
+{
+  "segments": [
+    {
+      "id": 1,
+      "speaker": 1,
+      "text": "Hello world, how are",
+      "translation": "",
+      "buffer": {
+        "transcription": "",
+        "diarization": " you on",
+        "translation": "Bonjour le monde"
+      }
+    }
+  ]
+}
+
+
+// ==== Frontend ====
+// <SPEAKER>1</SPEAKER>
+// <TRANSCRIPTION>Hello world, how are <DIARIZATION BUFFER> you on</DIARIZATION BUFFER></TRANSCRIPTION>
+// <TRANSLATION><TRANSLATION BUFFER>Bonjour le monde</TRANSLATION BUFFER></TRANSLATION>
+
+
+// Update 2
+{
+  "segments": [
+    {
+      "id": 1,
+      "speaker": 1,
+      "text": " you on this",
+      "translation": "Bonjour tout le monde",
+      "buffer": {
+        "transcription": "",
+        "diarization": " beautiful day",
+        "translation": ",comment"
+      }
+    },
+  ]
+}
+
+
+// ==== Frontend ====
+// <SPEAKER>1</SPEAKER>
+// <TRANSCRIPTION>Hello world, how are you on this<DIARIZATION BUFFER>  beautiful day</DIARIZATION BUFFER></TRANSCRIPTION>
+// <TRANSLATION>Bonjour tout le monde<TRANSLATION BUFFER>, comment</TRANSLATION BUFFER><TRANSLATION>
+```
+
+### Silence Segments
+
+Silence is represented with the speaker id = `-2`:
+
+```jsonc
+{
+  "id": 5,
+  "speaker": -2,
+  "text": "",
+  "start": 10.5,
+  "end": 12.3
+}
+```

docs/available_models.md

@@ -1,4 +1,4 @@
-# Available model sizes:
+# Available Whisper model sizes:
 
 - tiny.en (english only)
 - tiny
@@ -70,4 +70,40 @@
 2. Limited resources or need speed? → `small` or smaller
 3. Good hardware and want best quality? → `large-v3`
 4. Need fast, high-quality transcription without translation? → `large-v3-turbo`
-5. Need translation capabilities? → `large-v2` or `large-v3` (avoid turbo)
+5. Need translation capabilities? → `large-v2` or `large-v3` (avoid turbo)
+
+
+_______________________
+
+# Translation Models and Backend
+
+**Language Support**: ~200 languages
+
+## Distilled Model Sizes Available
+
+| Model | Size | Parameters | VRAM (FP16) | VRAM (INT8) | Quality |
+|-------|------|------------|-------------|-------------|---------|
+| 600M | 2.46 GB | 600M | ~1.5GB | ~800MB | Good, understandable |
+| 1.3B | 5.48 GB | 1.3B | ~3GB | ~1.5GB | Better accuracy, context |
+
+**Quality Impact**: 1.3B has ~15-25% better BLEU scores vs 600M across language pairs.
+
+## Backend Performance
+
+| Backend | Speed vs Base | Memory Usage | Quality Loss |
+|---------|---------------|--------------|--------------|
+| CTranslate2 | 6-10x faster | 40-60% less | ~5% BLEU drop |
+| Transformers | Baseline | High | None |
+| Transformers + MPS (on Apple Silicon) | 2x faster | Medium | None |
+
+**Metrics**:
+- CTranslate2: 50-100+ tokens/sec
+- Transformers: 10-30 tokens/sec
+- Apple Silicon with MPS: Up to 2x faster than CTranslate2
+
+## Quick Decision Matrix
+
+**Choose 600M**: Limited resources, close to 0 lag
+**Choose 1.3B**: Quality matters
+**Choose Transformers**: On Apple Silicon
+

docs/models_compatible_formats.md

@@ -0,0 +1,14 @@
+# Model Path Formats
+
+The `--model-path` parameter accepts:
+
+## File Path
+- **`.pt` format only** (required for AlignAtt policy decoder)
+
+## Directory Path (recommended)
+Must contain:
+- **`.pt` file** (required for decoder)
+
+May optionally contain:
+- **`.bin` file** - faster-whisper model for encoder (requires faster-whisper)
+- **`weights.npz`** or **`weights.safetensors`** - for encoder (requires whisper-mlx)

docs/supported_languages.md

@@ -0,0 +1,265 @@
+# Supported Languages
+
+WhisperLiveKit supports translation into **201 languages** from the FLORES-200 dataset through the NLLB (No Language Left Behind) translation system. 
+
+## How to Specify Languages
+
+You can specify languages in **three different ways**:
+
+1. **Language Name** (case-insensitive): `"English"`, `"French"`, `"Spanish"`
+2. **ISO Language Code**: `"en"`, `"fr"`, `"es"`
+3. **NLLB Code** (FLORES-200): `"eng_Latn"`, `"fra_Latn"`, `"spa_Latn"`
+
+## Usage Examples
+
+### Command Line
+```bash
+# Using language name
+whisperlivekit-server --target-language "French"
+
+# Using ISO code
+whisperlivekit-server --target-language fr
+
+# Using NLLB code
+whisperlivekit-server --target-language fra_Latn
+```
+
+### Python API
+```python
+from nllw.translation import get_language_info
+
+# Get language information by name
+lang_info = get_language_info("French")
+print(lang_info)
+# {'name': 'French', 'nllb': 'fra_Latn', 'language_code': 'fr'}
+
+# Get language information by ISO code
+lang_info = get_language_info("fr")
+
+# Get language information by NLLB code
+lang_info = get_language_info("fra_Latn")
+
+# All three return the same result
+```
+
+## Complete Language List
+
+The following table lists all 201 supported languages with their corresponding codes:
+
+| Language Name | ISO Code | NLLB Code |
+|---------------|----------|-----------|
+| Acehnese (Arabic script) | ace_Arab | ace_Arab |
+| Acehnese (Latin script) | ace_Latn | ace_Latn |
+| Mesopotamian Arabic | acm_Arab | acm_Arab |
+| Ta'izzi-Adeni Arabic | acq_Arab | acq_Arab |
+| Tunisian Arabic | aeb_Arab | aeb_Arab |
+| Afrikaans | af | afr_Latn |
+| South Levantine Arabic | ajp_Arab | ajp_Arab |
+| Akan | ak | aka_Latn |
+| Tosk Albanian | als | als_Latn |
+| Amharic | am | amh_Ethi |
+| North Levantine Arabic | apc_Arab | apc_Arab |
+| Modern Standard Arabic | ar | arb_Arab |
+| Modern Standard Arabic (Romanized) | arb_Latn | arb_Latn |
+| Najdi Arabic | ars_Arab | ars_Arab |
+| Moroccan Arabic | ary_Arab | ary_Arab |
+| Egyptian Arabic | arz_Arab | arz_Arab |
+| Assamese | as | asm_Beng |
+| Asturian | ast | ast_Latn |
+| Awadhi | awa | awa_Deva |
+| Central Aymara | ay | ayr_Latn |
+| South Azerbaijani | azb | azb_Arab |
+| North Azerbaijani | az | azj_Latn |
+| Bashkir | ba | bak_Cyrl |
+| Bambara | bm | bam_Latn |
+| Balinese | ban | ban_Latn |
+| Belarusian | be | bel_Cyrl |
+| Bemba | bem | bem_Latn |
+| Bengali | bn | ben_Beng |
+| Bhojpuri | bho | bho_Deva |
+| Banjar (Arabic script) | bjn_Arab | bjn_Arab |
+| Banjar (Latin script) | bjn_Latn | bjn_Latn |
+| Standard Tibetan | bo | bod_Tibt |
+| Bosnian | bs | bos_Latn |
+| Buginese | bug | bug_Latn |
+| Bulgarian | bg | bul_Cyrl |
+| Catalan | ca | cat_Latn |
+| Cebuano | ceb | ceb_Latn |
+| Czech | cs | ces_Latn |
+| Chokwe | cjk | cjk_Latn |
+| Central Kurdish | ckb | ckb_Arab |
+| Crimean Tatar | crh | crh_Latn |
+| Welsh | cy | cym_Latn |
+| Danish | da | dan_Latn |
+| German | de | deu_Latn |
+| Southwestern Dinka | dik | dik_Latn |
+| Dyula | dyu | dyu_Latn |
+| Dzongkha | dz | dzo_Tibt |
+| Greek | el | ell_Grek |
+| English | en | eng_Latn |
+| Esperanto | eo | epo_Latn |
+| Estonian | et | est_Latn |
+| Basque | eu | eus_Latn |
+| Ewe | ee | ewe_Latn |
+| Faroese | fo | fao_Latn |
+| Fijian | fj | fij_Latn |
+| Finnish | fi | fin_Latn |
+| Fon | fon | fon_Latn |
+| French | fr | fra_Latn |
+| Friulian | fur-IT | fur_Latn |
+| Nigerian Fulfulde | fuv | fuv_Latn |
+| West Central Oromo | om | gaz_Latn |
+| Scottish Gaelic | gd | gla_Latn |
+| Irish | ga-IE | gle_Latn |
+| Galician | gl | glg_Latn |
+| Guarani | gn | grn_Latn |
+| Gujarati | gu-IN | guj_Gujr |
+| Haitian Creole | ht | hat_Latn |
+| Hausa | ha | hau_Latn |
+| Hebrew | he | heb_Hebr |
+| Hindi | hi | hin_Deva |
+| Chhattisgarhi | hne | hne_Deva |
+| Croatian | hr | hrv_Latn |
+| Hungarian | hu | hun_Latn |
+| Armenian | hy-AM | hye_Armn |
+| Igbo | ig | ibo_Latn |
+| Ilocano | ilo | ilo_Latn |
+| Indonesian | id | ind_Latn |
+| Icelandic | is | isl_Latn |
+| Italian | it | ita_Latn |
+| Javanese | jv | jav_Latn |
+| Japanese | ja | jpn_Jpan |
+| Kabyle | kab | kab_Latn |
+| Jingpho | kac | kac_Latn |
+| Kamba | kam | kam_Latn |
+| Kannada | kn | kan_Knda |
+| Kashmiri (Arabic script) | kas_Arab | kas_Arab |
+| Kashmiri (Devanagari script) | kas_Deva | kas_Deva |
+| Georgian | ka | kat_Geor |
+| Kazakh | kk | kaz_Cyrl |
+| Kabiyè | kbp | kbp_Latn |
+| Kabuverdianu | kea | kea_Latn |
+| Halh Mongolian | mn | khk_Cyrl |
+| Khmer | km | khm_Khmr |
+| Kikuyu | ki | kik_Latn |
+| Kinyarwanda | rw | kin_Latn |
+| Kyrgyz | ky | kir_Cyrl |
+| Kimbundu | kmb | kmb_Latn |
+| Northern Kurdish | kmr | kmr_Latn |
+| Central Kanuri (Arabic script) | knc_Arab | knc_Arab |
+| Central Kanuri (Latin script) | knc_Latn | knc_Latn |
+| Kikongo | kg | kon_Latn |
+| Korean | ko | kor_Hang |
+| Lao | lo | lao_Laoo |
+| Ligurian | lij | lij_Latn |
+| Limburgish | li | lim_Latn |
+| Lingala | ln | lin_Latn |
+| Lithuanian | lt | lit_Latn |
+| Lombard | lmo | lmo_Latn |
+| Latgalian | ltg | ltg_Latn |
+| Luxembourgish | lb | ltz_Latn |
+| Luba-Kasai | lua | lua_Latn |
+| Ganda | lg | lug_Latn |
+| Luo | luo | luo_Latn |
+| Mizo | lus | lus_Latn |
+| Standard Latvian | lv | lvs_Latn |
+| Magahi | mag | mag_Deva |
+| Maithili | mai | mai_Deva |
+| Malayalam | ml-IN | mal_Mlym |
+| Marathi | mr | mar_Deva |
+| Minangkabau (Arabic script) | min_Arab | min_Arab |
+| Minangkabau (Latin script) | min_Latn | min_Latn |
+| Macedonian | mk | mkd_Cyrl |
+| Maltese | mt | mlt_Latn |
+| Meitei (Bengali script) | mni | mni_Beng |
+| Mossi | mos | mos_Latn |
+| Maori | mi | mri_Latn |
+| Burmese | my | mya_Mymr |
+| Dutch | nl | nld_Latn |
+| Norwegian Nynorsk | nn-NO | nno_Latn |
+| Norwegian Bokmål | nb | nob_Latn |
+| Nepali | ne-NP | npi_Deva |
+| Northern Sotho | nso | nso_Latn |
+| Nuer | nus | nus_Latn |
+| Nyanja | ny | nya_Latn |
+| Occitan | oc | oci_Latn |
+| Odia | or | ory_Orya |
+| Pangasinan | pag | pag_Latn |
+| Eastern Panjabi | pa | pan_Guru |
+| Papiamento | pap | pap_Latn |
+| Southern Pashto | pbt | pbt_Arab |
+| Western Persian | fa | pes_Arab |
+| Plateau Malagasy | mg | plt_Latn |
+| Polish | pl | pol_Latn |
+| Portuguese | pt-PT | por_Latn |
+| Dari | fa-AF | prs_Arab |
+| Ayacucho Quechua | qu | quy_Latn |
+| Romanian | ro | ron_Latn |
+| Rundi | rn | run_Latn |
+| Russian | ru | rus_Cyrl |
+| Sango | sg | sag_Latn |
+| Sanskrit | sa | san_Deva |
+| Santali | sat | sat_Olck |
+| Sicilian | scn | scn_Latn |
+| Shan | shn | shn_Mymr |
+| Sinhala | si-LK | sin_Sinh |
+| Slovak | sk | slk_Latn |
+| Slovenian | sl | slv_Latn |
+| Samoan | sm | smo_Latn |
+| Shona | sn | sna_Latn |
+| Sindhi | sd | snd_Arab |
+| Somali | so | som_Latn |
+| Southern Sotho | st | sot_Latn |
+| Spanish | es-ES | spa_Latn |
+| Sardinian | sc | srd_Latn |
+| Serbian | sr | srp_Cyrl |
+| Swati | ss | ssw_Latn |
+| Sundanese | su | sun_Latn |
+| Swedish | sv-SE | swe_Latn |
+| Swahili | sw | swh_Latn |
+| Silesian | szl | szl_Latn |
+| Tamil | ta | tam_Taml |
+| Tamasheq (Latin script) | taq_Latn | taq_Latn |
+| Tamasheq (Tifinagh script) | taq_Tfng | taq_Tfng |
+| Tatar | tt-RU | tat_Cyrl |
+| Telugu | te | tel_Telu |
+| Tajik | tg | tgk_Cyrl |
+| Tagalog | tl | tgl_Latn |
+| Thai | th | tha_Thai |
+| Tigrinya | ti | tir_Ethi |
+| Tok Pisin | tpi | tpi_Latn |
+| Tswana | tn | tsn_Latn |
+| Tsonga | ts | tso_Latn |
+| Turkmen | tk | tuk_Latn |
+| Tumbuka | tum | tum_Latn |
+| Turkish | tr | tur_Latn |
+| Twi | tw | twi_Latn |
+| Central Atlas Tamazight | tzm | tzm_Tfng |
+| Uyghur | ug | uig_Arab |
+| Ukrainian | uk | ukr_Cyrl |
+| Umbundu | umb | umb_Latn |
+| Urdu | ur | urd_Arab |
+| Northern Uzbek | uz | uzn_Latn |
+| Venetian | vec | vec_Latn |
+| Vietnamese | vi | vie_Latn |
+| Waray | war | war_Latn |
+| Wolof | wo | wol_Latn |
+| Xhosa | xh | xho_Latn |
+| Eastern Yiddish | yi | ydd_Hebr |
+| Yoruba | yo | yor_Latn |
+| Yue Chinese | yue | yue_Hant |
+| Chinese (Simplified) | zh-CN | zho_Hans |
+| Chinese (Traditional) | zh-TW | zho_Hant |
+| Standard Malay | ms | zsm_Latn |
+| Zulu | zu | zul_Latn |
+
+## Special Features
+
+### Multiple Script Support
+Several languages are available in multiple scripts (e.g., Arabic and Latin):
+- **Acehnese**: Arabic (`ace_Arab`) and Latin (`ace_Latn`)
+- **Banjar**: Arabic (`bjn_Arab`) and Latin (`bjn_Latn`)
+- **Kashmiri**: Arabic (`kas_Arab`) and Devanagari (`kas_Deva`)
+- **Minangkabau**: Arabic (`min_Arab`) and Latin (`min_Latn`)
+- **Tamasheq**: Latin (`taq_Latn`) and Tifinagh (`taq_Tfng`)
+- **Central Kanuri**: Arabic (`knc_Arab`) and Latin (`knc_Latn`)

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "whisperlivekit"
-version = "0.2.9"
+version = "0.2.13"
 description = "Real-time speech-to-text with speaker diarization using Whisper"
 readme = "README.md"
 authors = [
@@ -41,7 +41,8 @@ dependencies = [
 ]
 
 [project.optional-dependencies]
-sentence = ["mosestokenizer", "wtpsplit"]
+translation = ["nllw"]
+sentence_tokenizer = ["mosestokenizer", "wtpsplit"]
 
 [project.urls]
 Homepage = "https://github.com/QuentinFuxa/WhisperLiveKit"
@@ -50,8 +51,19 @@ Homepage = "https://github.com/QuentinFuxa/WhisperLiveKit"
 whisperlivekit-server = "whisperlivekit.basic_server:main"
 
 [tool.setuptools]
-packages = ["whisperlivekit", "whisperlivekit.diarization", "whisperlivekit.simul_whisper", "whisperlivekit.simul_whisper.whisper", "whisperlivekit.simul_whisper.whisper.assets", "whisperlivekit.simul_whisper.whisper.normalizers", "whisperlivekit.web", "whisperlivekit.whisper_streaming_custom"]
+packages = [
+    "whisperlivekit",
+    "whisperlivekit.diarization",
+    "whisperlivekit.simul_whisper",
+    "whisperlivekit.simul_whisper.whisper",
+    "whisperlivekit.simul_whisper.whisper.assets",
+    "whisperlivekit.simul_whisper.whisper.normalizers",
+    "whisperlivekit.web",
+    "whisperlivekit.whisper_streaming_custom",
+    "whisperlivekit.vad_models"
+]
 
 [tool.setuptools.package-data]
 whisperlivekit = ["web/*.html", "web/*.css", "web/*.js", "web/src/*.svg"]
 "whisperlivekit.simul_whisper.whisper.assets" = ["*.tiktoken", "*.npz"]
+"whisperlivekit.vad_models" = ["*.jit", "*.onnx"]

sync_extension.py

@@ -0,0 +1,38 @@
+import shutil
+import os
+from pathlib import Path
+
+def sync_extension_files():
+    """Copy core files from web directory to Chrome extension directory."""
+    
+    web_dir = Path("whisperlivekit/web")
+    extension_dir = Path("chrome-extension")
+    
+    files_to_sync = [
+        "live_transcription.html", "live_transcription.js", "live_transcription.css"
+    ]
+
+    svg_files = [
+        "system_mode.svg",
+        "light_mode.svg", 
+        "dark_mode.svg",
+        "settings.svg"
+    ]
+        
+    for file in files_to_sync:
+        src_path = web_dir / file
+        dest_path = extension_dir / file
+        
+        dest_path.parent.mkdir(parents=True, exist_ok=True)
+        shutil.copy2(src_path, dest_path)
+    
+    for svg_file in svg_files:
+        src_path = web_dir / "src" / svg_file
+        dest_path = extension_dir / "web" / "src" / svg_file
+        dest_path.parent.mkdir(parents=True, exist_ok=True)
+        shutil.copy2(src_path, dest_path)
+    
+
+if __name__ == "__main__":
+
+    sync_extension_files()

whisperlivekit/audio_processor.py

@@ -4,18 +4,30 @@ from time import time, sleep
 import math
 import logging
 import traceback
-from whisperlivekit.timed_objects import ASRToken, Silence, Line
+from whisperlivekit.timed_objects import ASRToken, Silence, Line, FrontData, State, Transcript, ChangeSpeaker
 from whisperlivekit.core import TranscriptionEngine, online_factory, online_diarization_factory, online_translation_factory
-from whisperlivekit.ffmpeg_manager import FFmpegManager, FFmpegState
 from whisperlivekit.silero_vad_iterator import FixedVADIterator
 from whisperlivekit.results_formater import format_output
-# Set up logging once
+from whisperlivekit.ffmpeg_manager import FFmpegManager, FFmpegState
+
 logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s")
 logger = logging.getLogger(__name__)
 logger.setLevel(logging.DEBUG)
 
 SENTINEL = object() # unique sentinel object for end of stream marker
 
+def cut_at(cumulative_pcm, cut_sec):
+    cumulative_len = 0
+    cut_sample = int(cut_sec * 16000)
+    
+    for ind, pcm_array in enumerate(cumulative_pcm):
+        if (cumulative_len + len(pcm_array)) >= cut_sample:
+            cut_chunk = cut_sample - cumulative_len
+            before = np.concatenate(cumulative_pcm[:ind] + [cumulative_pcm[ind][:cut_chunk]])
+            after = [cumulative_pcm[ind][cut_chunk:]] + cumulative_pcm[ind+1:]
+            return before, after
+        cumulative_len += len(pcm_array)
+    return np.concatenate(cumulative_pcm), []
 
 async def get_all_from_queue(queue):
     items = []
@@ -49,48 +61,50 @@ class AudioProcessor:
         self.bytes_per_sample = 2
         self.bytes_per_sec = self.samples_per_sec * self.bytes_per_sample
         self.max_bytes_per_sec = 32000 * 5  # 5 seconds of audio at 32 kHz
-        self.last_ffmpeg_activity = time()
-        self.ffmpeg_health_check_interval = 5
-        self.ffmpeg_max_idle_time = 10
         self.is_pcm_input = self.args.pcm_input
-        self.debug = False
 
         # State management
         self.is_stopping = False
         self.silence = False
         self.silence_duration = 0.0
-        self.tokens = []
-        self.translated_segments = []
-        self.buffer_transcription = ""
-        self.buffer_diarization = ""
-        self.end_buffer = 0
-        self.end_attributed_speaker = 0
+        self.state = State()
         self.lock = asyncio.Lock()
-        self.beg_loop = None #to deal with a potential little lag at the websocket initialization, this is now set in process_audio
         self.sep = " "  # Default separator
-        self.last_response_content = ""
+        self.last_response_content = FrontData()
+        self.last_detected_speaker = None
+        self.speaker_languages = {}
+        self.diarization_before_transcription = False
+
+        self.segments = []
+        
+
+        if self.diarization_before_transcription:
+            self.cumulative_pcm = []
+            self.last_start = 0.0
+            self.last_end = 0.0
         
         # Models and processing
         self.asr = models.asr
-        self.tokenizer = models.tokenizer
         self.vac_model = models.vac_model
         if self.args.vac:
             self.vac = FixedVADIterator(models.vac_model)
         else:
             self.vac = None
-            
-        self.ffmpeg_manager = FFmpegManager(
-            sample_rate=self.sample_rate,
-            channels=self.channels
-        )
-        
-        async def handle_ffmpeg_error(error_type: str):
-            logger.error(f"FFmpeg error: {error_type}")
-            self._ffmpeg_error = error_type
-        
-        self.ffmpeg_manager.on_error_callback = handle_ffmpeg_error
+                         
+        self.ffmpeg_manager = None
+        self.ffmpeg_reader_task = None
         self._ffmpeg_error = None
-        
+
+        if not self.is_pcm_input:
+            self.ffmpeg_manager = FFmpegManager(
+                sample_rate=self.sample_rate,
+                channels=self.channels
+            )
+            async def handle_ffmpeg_error(error_type: str):
+                logger.error(f"FFmpeg error: {error_type}")
+                self._ffmpeg_error = error_type
+            self.ffmpeg_manager.on_error_callback = handle_ffmpeg_error
+             
         self.transcription_queue = asyncio.Queue() if self.args.transcription else None
         self.diarization_queue = asyncio.Queue() if self.args.diarization else None
         self.translation_queue = asyncio.Queue() if self.args.target_language else None
@@ -98,41 +112,31 @@ class AudioProcessor:
 
         self.transcription_task = None
         self.diarization_task = None
-        self.ffmpeg_reader_task = None
+        self.translation_task = None
         self.watchdog_task = None
         self.all_tasks_for_cleanup = []
         
+        self.transcription = None
+        self.translation = None
+        self.diarization = None
+
         if self.args.transcription:
-            self.online = online_factory(self.args, models.asr, models.tokenizer)            
+            self.transcription = online_factory(self.args, models.asr)        
+            self.sep = self.transcription.asr.sep   
         if self.args.diarization:
             self.diarization = online_diarization_factory(self.args, models.diarization_model)
-        if self.args.target_language:
-            self.online_translation = online_translation_factory(self.args, models.translation_model)
+        if models.translation_model:
+            self.translation = online_translation_factory(self.args, models.translation_model)
 
     def convert_pcm_to_float(self, pcm_buffer):
         """Convert PCM buffer in s16le format to normalized NumPy array."""
         return np.frombuffer(pcm_buffer, dtype=np.int16).astype(np.float32) / 32768.0
 
-    async def update_transcription(self, new_tokens, buffer, end_buffer, sep):
-        """Thread-safe update of transcription with new data."""
-        async with self.lock:
-            self.tokens.extend(new_tokens)
-            self.buffer_transcription = buffer
-            self.end_buffer = end_buffer
-            self.sep = sep
-            
-    async def update_diarization(self, end_attributed_speaker, buffer_diarization=""):
-        """Thread-safe update of diarization with new data."""
-        async with self.lock:
-            self.end_attributed_speaker = end_attributed_speaker
-            if buffer_diarization:
-                self.buffer_diarization = buffer_diarization
-            
     async def add_dummy_token(self):
         """Placeholder token when no transcription is available."""
         async with self.lock:
-            current_time = time() - self.beg_loop if self.beg_loop else 0
-            self.tokens.append(ASRToken(
+            current_time = time() - self.state.beg_loop
+            self.state.tokens.append(ASRToken(
                 start=current_time, end=current_time + 1,
                 text=".", speaker=-1, is_dummy=True
             ))
@@ -142,45 +146,30 @@ class AudioProcessor:
         async with self.lock:
             current_time = time()
             
-            # Calculate remaining times
             remaining_transcription = 0
-            if self.end_buffer > 0:
-                remaining_transcription = max(0, round(current_time - self.beg_loop - self.end_buffer, 1))
+            if self.state.end_buffer > 0:
+                remaining_transcription = max(0, round(current_time - self.state.beg_loop - self.state.end_buffer, 1))
                 
             remaining_diarization = 0
-            if self.tokens:
-                latest_end = max(self.end_buffer, self.tokens[-1].end if self.tokens else 0)
-                remaining_diarization = max(0, round(latest_end - self.end_attributed_speaker, 1))
+            if self.state.tokens:
+                latest_end = max(self.state.end_buffer, self.state.tokens[-1].end if self.state.tokens else 0)
+                remaining_diarization = max(0, round(latest_end - self.state.end_attributed_speaker, 1))
                 
-            return {
-                "tokens": self.tokens.copy(),
-                "translated_segments": self.translated_segments.copy(),
-                "buffer_transcription": self.buffer_transcription,
-                "buffer_diarization": self.buffer_diarization,
-                "end_buffer": self.end_buffer,
-                "end_attributed_speaker": self.end_attributed_speaker,
-                "sep": self.sep,
-                "remaining_time_transcription": remaining_transcription,
-                "remaining_time_diarization": remaining_diarization
-            }
+            self.state.remaining_time_transcription = remaining_transcription
+            self.state.remaining_time_diarization = remaining_diarization
             
-    async def reset(self):
-        """Reset all state variables to initial values."""
-        async with self.lock:
-            self.tokens = []
-            self.translated_segments = []
-            self.buffer_transcription = self.buffer_diarization = ""
-            self.end_buffer = self.end_attributed_speaker = 0
-            self.beg_loop = time()
+            return self.state
 
     async def ffmpeg_stdout_reader(self):
-        """Read audio data from FFmpeg stdout and process it."""
+        """Read audio data from FFmpeg stdout and process it into the PCM pipeline."""
         beg = time()
-        
         while True:
             try:
-                # Check if FFmpeg is running
-                state = await self.ffmpeg_manager.get_state()
+                if self.is_stopping:
+                    logger.info("Stopping ffmpeg_stdout_reader due to stopping flag.")
+                    break
+
+                state = await self.ffmpeg_manager.get_state() if self.ffmpeg_manager else FFmpegState.STOPPED
                 if state == FFmpegState.FAILED:
                     logger.error("FFmpeg is in FAILED state, cannot read data")
                     break
@@ -188,55 +177,41 @@ class AudioProcessor:
                     logger.info("FFmpeg is stopped")
                     break
                 elif state != FFmpegState.RUNNING:
-                    logger.warning(f"FFmpeg is in {state} state, waiting...")
-                    await asyncio.sleep(0.5)
+                    await asyncio.sleep(0.1)
                     continue
-                
+
                 current_time = time()
-                elapsed_time = math.floor((current_time - beg) * 10) / 10
-                buffer_size = max(int(32000 * elapsed_time), 4096)
+                elapsed_time = max(0.0, current_time - beg)
+                buffer_size = max(int(32000 * elapsed_time), 4096)  # dynamic read
                 beg = current_time
 
                 chunk = await self.ffmpeg_manager.read_data(buffer_size)
-                        
                 if not chunk:
-                    if self.is_stopping:
-                        logger.info("FFmpeg stdout closed, stopping.")
-                        break
-                    else:
-                        # No data available, but not stopping - FFmpeg might be restarting
-                        await asyncio.sleep(0.1)
-                        continue
-                    
+                    # No data currently available
+                    await asyncio.sleep(0.05)
+                    continue
+
                 self.pcm_buffer.extend(chunk)
                 await self.handle_pcm_data()
-                    
-                    
-                    
+
+            except asyncio.CancelledError:
+                logger.info("ffmpeg_stdout_reader cancelled.")
+                break
             except Exception as e:
                 logger.warning(f"Exception in ffmpeg_stdout_reader: {e}")
-                logger.warning(f"Traceback: {traceback.format_exc()}")
-                # Try to recover by waiting a bit
-                await asyncio.sleep(1)
-                
-                # Check if we should exit
-                if self.is_stopping:
-                    break
-        
-        logger.info("FFmpeg stdout processing finished. Signaling downstream processors.")
-        if self.args.transcription and self.transcription_queue:
-            await self.transcription_queue.put(SENTINEL)
-            logger.debug("Sentinel put into transcription_queue.")
-        if self.args.diarization and self.diarization_queue:
-            await self.diarization_queue.put(SENTINEL)
-            logger.debug("Sentinel put into diarization_queue.")
-        if self.args.target_language and self.translation_queue:
-            await self.translation_queue.put(SENTINEL)
+                logger.debug(f"Traceback: {traceback.format_exc()}")
+                await asyncio.sleep(0.2)
 
+        logger.info("FFmpeg stdout processing finished. Signaling downstream processors if needed.")
+        if not self.diarization_before_transcription and self.transcription_queue:
+            await self.transcription_queue.put(SENTINEL)
+        if self.diarization:
+            await self.diarization_queue.put(SENTINEL)
+        if self.translation:
+            await self.translation_queue.put(SENTINEL)
 
     async def transcription_processor(self):
         """Process audio chunks for transcription."""
-        self.sep = self.online.asr.sep
         cumulative_pcm_duration_stream_time = 0.0
         
         while True:
@@ -246,64 +221,56 @@ class AudioProcessor:
                     logger.debug("Transcription processor received sentinel. Finishing.")
                     self.transcription_queue.task_done()
                     break
-                
-                if not self.online:
-                    logger.warning("Transcription processor: self.online not initialized.")
-                    self.transcription_queue.task_done()
-                    continue
 
-                asr_internal_buffer_duration_s = len(getattr(self.online, 'audio_buffer', [])) / self.online.SAMPLING_RATE
-                transcription_lag_s = max(0.0, time() - self.beg_loop - self.end_buffer)
+                asr_internal_buffer_duration_s = len(getattr(self.transcription, 'audio_buffer', [])) / self.transcription.SAMPLING_RATE
+                transcription_lag_s = max(0.0, time() - self.state.beg_loop - self.state.end_buffer)
                 asr_processing_logs = f"internal_buffer={asr_internal_buffer_duration_s:.2f}s | lag={transcription_lag_s:.2f}s |"
                 if type(item) is Silence:
                     asr_processing_logs += f" + Silence of = {item.duration:.2f}s"
-                    if self.tokens:
-                        asr_processing_logs += f" | last_end = {self.tokens[-1].end} |"
-                logger.info(asr_processing_logs)
-                
-                if type(item) is Silence:
+                    if self.state.tokens:
+                        asr_processing_logs += f" | last_end = {self.state.tokens[-1].end} |"
+                    logger.info(asr_processing_logs)
                     cumulative_pcm_duration_stream_time += item.duration
-                    self.online.insert_silence(item.duration, self.tokens[-1].end if self.tokens else 0)
+                    self.transcription.insert_silence(item.duration, self.state.tokens[-1].end if self.state.tokens else 0)
                     continue
-                
-                if isinstance(item, np.ndarray):
+                elif isinstance(item, ChangeSpeaker):
+                    self.transcription.new_speaker(item)
+                elif isinstance(item, np.ndarray):
                     pcm_array = item
-                else:
-                    raise Exception('item should be pcm_array')
+                
+                logger.info(asr_processing_logs)
                 
                 duration_this_chunk = len(pcm_array) / self.sample_rate
                 cumulative_pcm_duration_stream_time += duration_this_chunk
                 stream_time_end_of_current_pcm = cumulative_pcm_duration_stream_time
 
-                self.online.insert_audio_chunk(pcm_array, stream_time_end_of_current_pcm)
-                new_tokens, current_audio_processed_upto = self.online.process_iter()
+                self.transcription.insert_audio_chunk(pcm_array, stream_time_end_of_current_pcm)
+                new_tokens, current_audio_processed_upto = await asyncio.to_thread(self.transcription.process_iter)
                 
-                # Get buffer information
-                _buffer_transcript_obj = self.online.get_buffer()
-                buffer_text = _buffer_transcript_obj.text
+                _buffer_transcript = self.transcription.get_buffer()
+                buffer_text = _buffer_transcript.text
 
                 if new_tokens:
                     validated_text = self.sep.join([t.text for t in new_tokens])
                     if buffer_text.startswith(validated_text):
-                        buffer_text = buffer_text[len(validated_text):].lstrip()
+                        _buffer_transcript.text = buffer_text[len(validated_text):].lstrip()
 
-                candidate_end_times = [self.end_buffer]
+                candidate_end_times = [self.state.end_buffer]
 
                 if new_tokens:
                     candidate_end_times.append(new_tokens[-1].end)
                 
-                if _buffer_transcript_obj.end is not None:
-                    candidate_end_times.append(_buffer_transcript_obj.end)
+                if _buffer_transcript.end is not None:
+                    candidate_end_times.append(_buffer_transcript.end)
                 
                 candidate_end_times.append(current_audio_processed_upto)
                 
-                new_end_buffer = max(candidate_end_times)
+                async with self.lock:
+                    self.state.tokens.extend(new_tokens)
+                    self.state.buffer_transcription = _buffer_transcript
+                    self.state.end_buffer = max(candidate_end_times)
                 
-                await self.update_transcription(
-                    new_tokens, buffer_text, new_end_buffer, self.sep
-                )
-                
-                if new_tokens and self.args.target_language and self.translation_queue:
+                if self.translation_queue:
                     for token in new_tokens:
                         await self.translation_queue.put(token)
                         
@@ -314,13 +281,22 @@ class AudioProcessor:
                 logger.warning(f"Traceback: {traceback.format_exc()}")
                 if 'pcm_array' in locals() and pcm_array is not SENTINEL : # Check if pcm_array was assigned from queue
                     self.transcription_queue.task_done()
+        
+        if self.is_stopping:
+            logger.info("Transcription processor finishing due to stopping flag.")
+            if self.diarization_queue:
+                await self.diarization_queue.put(SENTINEL)
+            if self.translation_queue:
+                await self.translation_queue.put(SENTINEL)
+
         logger.info("Transcription processor task finished.")
 
 
     async def diarization_processor(self, diarization_obj):
         """Process audio chunks for speaker diarization."""
-        buffer_diarization = ""
-        cumulative_pcm_duration_stream_time = 0.0
+        if self.diarization_before_transcription:
+            self.current_speaker = 0
+            await self.transcription_queue.put(ChangeSpeaker(speaker=self.current_speaker, start=0.0))
         while True:
             try:
                 item = await self.diarization_queue.get()
@@ -328,30 +304,49 @@ class AudioProcessor:
                     logger.debug("Diarization processor received sentinel. Finishing.")
                     self.diarization_queue.task_done()
                     break
-                
-                if type(item) is Silence:
-                    cumulative_pcm_duration_stream_time += item.duration
+                elif type(item) is Silence:
                     diarization_obj.insert_silence(item.duration)
                     continue
-    
-                if isinstance(item, np.ndarray):
+                elif isinstance(item, np.ndarray):
                     pcm_array = item
                 else:
                     raise Exception('item should be pcm_array') 
                 
+                
+                
                 # Process diarization
                 await diarization_obj.diarize(pcm_array)
-                
-                async with self.lock:
-                    self.tokens = diarization_obj.assign_speakers_to_tokens(
-                        self.tokens,
-                        use_punctuation_split=self.args.punctuation_split
-                    )
-                    if len(self.tokens) > 0:
-                        self.end_attributed_speaker = max(self.tokens[-1].end, self.end_attributed_speaker)
-                    if buffer_diarization:
-                        self.buffer_diarization = buffer_diarization
-                
+                if self.diarization_before_transcription:
+                    segments = diarization_obj.get_segments()
+                    self.cumulative_pcm.append(pcm_array)
+                    if segments:
+                        last_segment = segments[-1]                    
+                        if last_segment.speaker != self.current_speaker:
+                            cut_sec = last_segment.start - self.last_end
+                            to_transcript, self.cumulative_pcm = cut_at(self.cumulative_pcm, cut_sec)
+                            await self.transcription_queue.put(to_transcript)
+                            
+                            self.current_speaker = last_segment.speaker
+                            await self.transcription_queue.put(ChangeSpeaker(speaker=self.current_speaker, start=last_segment.start))
+                            
+                            cut_sec = last_segment.end - last_segment.start
+                            to_transcript, self.cumulative_pcm = cut_at(self.cumulative_pcm, cut_sec)
+                            await self.transcription_queue.put(to_transcript)                            
+                            self.last_start = last_segment.start
+                            self.last_end = last_segment.end
+                        else:
+                            cut_sec = last_segment.end - self.last_end
+                            to_transcript, self.cumulative_pcm = cut_at(self.cumulative_pcm, cut_sec)
+                            await self.transcription_queue.put(to_transcript)
+                            self.last_end = last_segment.end
+                elif not self.diarization_before_transcription:           
+                    async with self.lock:
+                        self.state.tokens = diarization_obj.assign_speakers_to_tokens(
+                            self.state.tokens,
+                            use_punctuation_split=self.args.punctuation_split
+                        )
+                if len(self.state.tokens) > 0:
+                    self.state.end_attributed_speaker = max(self.state.tokens[-1].end, self.state.end_attributed_speaker)
                 self.diarization_queue.task_done()
                 
             except Exception as e:
@@ -361,20 +356,23 @@ class AudioProcessor:
                     self.diarization_queue.task_done()
         logger.info("Diarization processor task finished.")
 
-    async def translation_processor(self, online_translation):
+    async def translation_processor(self):
         # the idea is to ignore diarization for the moment. We use only transcription tokens. 
         # And the speaker is attributed given the segments used for the translation
         # in the future we want to have different languages for each speaker etc, so it will be more complex.
         while True:
             try:
-                token = await self.translation_queue.get() #block until at least 1 token
-                if token is SENTINEL:
+                item = await self.translation_queue.get() #block until at least 1 token
+                if item is SENTINEL:
                     logger.debug("Translation processor received sentinel. Finishing.")
                     self.translation_queue.task_done()
                     break
+                elif type(item) is Silence:
+                    self.translation.insert_silence(item.duration)
+                    continue
                 
                 # get all the available tokens for translation. The more words, the more precise
-                tokens_to_process = [token]
+                tokens_to_process = [item]
                 additional_tokens = await get_all_from_queue(self.translation_queue)
                 
                 sentinel_found = False
@@ -382,11 +380,17 @@ class AudioProcessor:
                     if additional_token is SENTINEL:
                         sentinel_found = True
                         break
-                    tokens_to_process.append(additional_token)                
+                    elif type(additional_token) is Silence:
+                        self.translation.insert_silence(additional_token.duration)
+                        continue
+                    else:
+                        tokens_to_process.append(additional_token)                
                 if tokens_to_process:
-                    online_translation.insert_tokens(tokens_to_process)
-                    self.translated_segments = online_translation.process()
-                
+                    self.translation.insert_tokens(tokens_to_process)
+                    translation_validated_segments, translation_buffer = await asyncio.to_thread(self.translation.process)
+                    async with self.lock:
+                        self.state.translation_validated_segments = translation_validated_segments
+                        self.state.translation_buffer = translation_buffer
                 self.translation_queue.task_done()
                 for _ in additional_tokens:
                     self.translation_queue.task_done()
@@ -398,7 +402,7 @@ class AudioProcessor:
             except Exception as e:
                 logger.warning(f"Exception in translation_processor: {e}")
                 logger.warning(f"Traceback: {traceback.format_exc()}")
-                if 'token' in locals() and token is not SENTINEL:
+                if 'token' in locals() and item is not SENTINEL:
                     self.translation_queue.task_done()
                 if 'additional_tokens' in locals():
                     for _ in additional_tokens:
@@ -407,152 +411,104 @@ class AudioProcessor:
 
     async def results_formatter(self):
         """Format processing results for output."""
-        last_sent_trans = None
-        last_sent_diar = None
         while True:
             try:
-                ffmpeg_state = await self.ffmpeg_manager.get_state()
-                if ffmpeg_state == FFmpegState.FAILED and self._ffmpeg_error:
-                    yield {
-                        "status": "error",
-                        "error": f"FFmpeg error: {self._ffmpeg_error}",
-                        "lines": [],
-                        "buffer_transcription": "",
-                        "buffer_diarization": "",
-                        "remaining_time_transcription": 0,
-                        "remaining_time_diarization": 0
-                    }
+                if self._ffmpeg_error:
+                    yield FrontData(status="error", error=f"FFmpeg error: {self._ffmpeg_error}")
                     self._ffmpeg_error = None
                     await asyncio.sleep(1)
                     continue
-                
-                # Get current state
+
                 state = await self.get_current_state()
-                tokens = state["tokens"]
-                buffer_transcription = state["buffer_transcription"]
-                buffer_diarization = state["buffer_diarization"]
-                end_attributed_speaker = state["end_attributed_speaker"]
-                sep = state["sep"]
-                                
-                # Add dummy tokens if needed
-                if (not tokens or tokens[-1].is_dummy) and not self.args.transcription and self.args.diarization:
-                    await self.add_dummy_token()
-                    sleep(0.5)
-                    state = await self.get_current_state()
-                    tokens = state["tokens"]
                 
-                # Format output
-                lines, undiarized_text, buffer_transcription, buffer_diarization = format_output(
+                lines, undiarized_text = format_output(
                     state,
                     self.silence,
-                    current_time = time() - self.beg_loop if self.beg_loop else None,
                     args = self.args,
-                    debug = self.debug
+                    sep=self.sep
                 )
-                # Handle undiarized text
+                if lines and lines[-1].speaker == -2:
+                    buffer_transcription = Transcript()
+                else:
+                    buffer_transcription = state.buffer_transcription
+
+                buffer_diarization = ''
                 if undiarized_text:
-                    combined = sep.join(undiarized_text)
-                    if buffer_transcription:
-                        combined += sep
-                    await self.update_diarization(end_attributed_speaker, combined)
-                    buffer_diarization = combined
+                    buffer_diarization = self.sep.join(undiarized_text)
+
+                    async with self.lock:
+                        self.state.end_attributed_speaker = state.end_attributed_speaker
                 
                 response_status = "active_transcription"
-                if not tokens and not buffer_transcription and not buffer_diarization:
+                if not state.tokens and not buffer_transcription and not buffer_diarization:
                     response_status = "no_audio_detected"
                     lines = []
-                elif response_status == "active_transcription" and not lines:
+                elif not lines:
                     lines = [Line(
                         speaker=1,
-                        start=state.get("end_buffer", 0),
-                        end=state.get("end_buffer", 0)        
+                        start=state.end_buffer,
+                        end=state.end_buffer
                     )]
                 
-                response = {
-                    "status": response_status,
-                    "lines": [line.to_dict() for line in lines],
-                    "buffer_transcription": buffer_transcription,
-                    "buffer_diarization": buffer_diarization,
-                    "remaining_time_transcription": state["remaining_time_transcription"],
-                    "remaining_time_diarization": state["remaining_time_diarization"] if self.args.diarization else 0
-                }
-                
-                current_response_signature = f"{response_status} | " + \
-                                           ' '.join([f"{line.speaker} {line.text}" for line in lines]) + \
-                                           f" | {buffer_transcription} | {buffer_diarization}"
-                
-                trans = state["remaining_time_transcription"]
-                diar = state["remaining_time_diarization"]
-                should_push = (
-                    current_response_signature != self.last_response_content
-                    or last_sent_trans is None
-                    or round(trans, 1) != round(last_sent_trans, 1)
-                    or round(diar, 1) != round(last_sent_diar, 1)
+                response = FrontData(
+                    status=response_status,
+                    lines=lines,
+                    buffer_transcription=buffer_transcription.text.strip(),
+                    buffer_diarization=buffer_diarization,
+                    remaining_time_transcription=state.remaining_time_transcription,
+                    remaining_time_diarization=state.remaining_time_diarization if self.args.diarization else 0
                 )
-                if should_push and (lines or buffer_transcription or buffer_diarization or response_status == "no_audio_detected" or trans > 0 or diar > 0):
+                                
+                should_push = (response != self.last_response_content)
+                if should_push and (lines or buffer_transcription or buffer_diarization or response_status == "no_audio_detected"):
                     yield response
-                    self.last_response_content = current_response_signature
-                    last_sent_trans = trans
-                    last_sent_diar = diar
+                    self.last_response_content = response
                 
-                # Check for termination condition
-                if self.is_stopping:
-                    all_processors_done = True
-                    if self.args.transcription and self.transcription_task and not self.transcription_task.done():
-                        all_processors_done = False
-                    if self.args.diarization and self.diarization_task and not self.diarization_task.done():
-                        all_processors_done = False
-                    
-                    if all_processors_done:
-                        logger.info("Results formatter: All upstream processors are done and in stopping state. Terminating.")
-                        return
+                if self.is_stopping and self.transcription_task and self.transcription_task.done() and self.diarization_task and self.diarization_task.done():
+                    logger.info("Results formatter: All upstream processors are done and in stopping state. Terminating.")
+                    return
                 
-                await asyncio.sleep(0.1)  # Avoid overwhelming the client
+                await asyncio.sleep(0.05)
                 
             except Exception as e:
-                logger.warning(f"Exception in results_formatter: {e}")
-                logger.warning(f"Traceback: {traceback.format_exc()}")
-                await asyncio.sleep(0.5)  # Back off on error
+                logger.warning(f"Exception in results_formatter. Traceback: {traceback.format_exc()}")
+                await asyncio.sleep(0.5)
         
     async def create_tasks(self):
         """Create and start processing tasks."""
         self.all_tasks_for_cleanup = []
         processing_tasks_for_watchdog = []
 
-        success = await self.ffmpeg_manager.start()
-        if not success:
-            logger.error("Failed to start FFmpeg manager")
-            async def error_generator():
-                yield {
-                    "status": "error", 
-                    "error": "FFmpeg failed to start. Please check that FFmpeg is installed.",
-                    "lines": [],
-                    "buffer_transcription": "",
-                    "buffer_diarization": "",
-                    "remaining_time_transcription": 0,
-                    "remaining_time_diarization": 0
-                }
-            return error_generator()
+        # If using FFmpeg (non-PCM input), start it and spawn stdout reader
+        if not self.is_pcm_input:
+            success = await self.ffmpeg_manager.start()
+            if not success:
+                logger.error("Failed to start FFmpeg manager")
+                async def error_generator():
+                    yield FrontData(
+                        status="error",
+                        error="FFmpeg failed to start. Please check that FFmpeg is installed."
+                    )
+                return error_generator()
+            self.ffmpeg_reader_task = asyncio.create_task(self.ffmpeg_stdout_reader())
+            self.all_tasks_for_cleanup.append(self.ffmpeg_reader_task)
+            processing_tasks_for_watchdog.append(self.ffmpeg_reader_task)
 
-        if self.args.transcription and self.online:
+        if self.transcription:
             self.transcription_task = asyncio.create_task(self.transcription_processor())
             self.all_tasks_for_cleanup.append(self.transcription_task)
             processing_tasks_for_watchdog.append(self.transcription_task)
             
-        if self.args.diarization and self.diarization:
+        if self.diarization:
             self.diarization_task = asyncio.create_task(self.diarization_processor(self.diarization))
             self.all_tasks_for_cleanup.append(self.diarization_task)
             processing_tasks_for_watchdog.append(self.diarization_task)
         
-        if self.args.target_language and self.args.lan != 'auto':
-            self.translation_task = asyncio.create_task(self.translation_processor(self.online_translation))
+        if self.translation:
+            self.translation_task = asyncio.create_task(self.translation_processor())
             self.all_tasks_for_cleanup.append(self.translation_task)
             processing_tasks_for_watchdog.append(self.translation_task)
         
-        self.ffmpeg_reader_task = asyncio.create_task(self.ffmpeg_stdout_reader())
-        self.all_tasks_for_cleanup.append(self.ffmpeg_reader_task)
-        processing_tasks_for_watchdog.append(self.ffmpeg_reader_task)
-
         # Monitor overall system health
         self.watchdog_task = asyncio.create_task(self.watchdog(processing_tasks_for_watchdog))
         self.all_tasks_for_cleanup.append(self.watchdog_task)
@@ -573,15 +529,6 @@ class AudioProcessor:
                             logger.error(f"{task_name} unexpectedly completed with exception: {exc}")
                         else:
                             logger.info(f"{task_name} completed normally.")
-                
-                # Check FFmpeg status through the manager
-                ffmpeg_state = await self.ffmpeg_manager.get_state()
-                if ffmpeg_state == FFmpegState.FAILED:
-                    logger.error("FFmpeg is in FAILED state, notifying results formatter")
-                    # FFmpeg manager will handle its own recovery
-                elif ffmpeg_state == FFmpegState.STOPPED and not self.is_stopping:
-                    logger.warning("FFmpeg unexpectedly stopped, attempting restart")
-                    await self.ffmpeg_manager.restart()
                     
             except asyncio.CancelledError:
                 logger.info("Watchdog task cancelled.")
@@ -597,28 +544,38 @@ class AudioProcessor:
             if task and not task.done():
                 task.cancel()
             
-            created_tasks = [t for t in self.all_tasks_for_cleanup if t]
-            if created_tasks:
-                await asyncio.gather(*created_tasks, return_exceptions=True)
-            logger.info("All processing tasks cancelled or finished.")
-            await self.ffmpeg_manager.stop()
-            logger.info("FFmpeg manager stopped.")
-            if self.args.diarization and hasattr(self, 'diarization') and hasattr(self.diarization, 'close'):
-                self.diarization.close()
-            logger.info("AudioProcessor cleanup complete.")
+        created_tasks = [t for t in self.all_tasks_for_cleanup if t]
+        if created_tasks:
+            await asyncio.gather(*created_tasks, return_exceptions=True)
+        logger.info("All processing tasks cancelled or finished.")
+
+        if not self.is_pcm_input and self.ffmpeg_manager:
+            try:
+                await self.ffmpeg_manager.stop()
+                logger.info("FFmpeg manager stopped.")
+            except Exception as e:
+                logger.warning(f"Error stopping FFmpeg manager: {e}")
+        if self.diarization:
+            self.diarization.close()
+        logger.info("AudioProcessor cleanup complete.")
 
 
     async def process_audio(self, message):
         """Process incoming audio data."""
 
-        if not self.beg_loop:
-            self.beg_loop = time()
+        if not self.state.beg_loop:
+            self.state.beg_loop = time()
 
         if not message:
             logger.info("Empty audio message received, initiating stop sequence.")
             self.is_stopping = True
-            # Signal FFmpeg manager to stop accepting data
-            await self.ffmpeg_manager.stop()
+             
+            if self.transcription_queue:
+                await self.transcription_queue.put(SENTINEL)
+
+            if not self.is_pcm_input and self.ffmpeg_manager:
+                await self.ffmpeg_manager.stop()
+
             return
 
         if self.is_stopping:
@@ -629,6 +586,9 @@ class AudioProcessor:
             self.pcm_buffer.extend(message)
             await self.handle_pcm_data()
         else:
+            if not self.ffmpeg_manager:
+                logger.error("FFmpeg manager not initialized for non-PCM input.")
+                return
             success = await self.ffmpeg_manager.write_data(message)
             if not success:
                 ffmpeg_state = await self.ffmpeg_manager.get_state()
@@ -648,9 +608,13 @@ class AudioProcessor:
                 f"Consider using a smaller model."
             )
 
-        # Process audio chunk
-        pcm_array = self.convert_pcm_to_float(self.pcm_buffer[:self.max_bytes_per_sec])
-        self.pcm_buffer = self.pcm_buffer[self.max_bytes_per_sec:]
+        chunk_size = min(len(self.pcm_buffer), self.max_bytes_per_sec)
+        aligned_chunk_size = (chunk_size // self.bytes_per_sample) * self.bytes_per_sample
+        
+        if aligned_chunk_size == 0:
+            return
+        pcm_array = self.convert_pcm_to_float(self.pcm_buffer[:aligned_chunk_size])
+        self.pcm_buffer = self.pcm_buffer[aligned_chunk_size:]
 
         res = None
         end_of_audio = False
@@ -667,13 +631,15 @@ class AudioProcessor:
                 silence_buffer = Silence(duration=time() - self.start_silence)
 
         if silence_buffer:
-            if self.args.transcription and self.transcription_queue:
+            if not self.diarization_before_transcription and self.transcription_queue:
                 await self.transcription_queue.put(silence_buffer)
             if self.args.diarization and self.diarization_queue:
                 await self.diarization_queue.put(silence_buffer)
+            if self.translation_queue:
+                await self.translation_queue.put(silence_buffer)
 
         if not self.silence:
-            if self.args.transcription and self.transcription_queue:
+            if not self.diarization_before_transcription and self.transcription_queue:
                 await self.transcription_queue.put(pcm_array.copy())
 
             if self.args.diarization and self.diarization_queue:

whisperlivekit/basic_server.py

@@ -5,9 +5,6 @@ from fastapi.middleware.cors import CORSMiddleware
 from whisperlivekit import TranscriptionEngine, AudioProcessor, get_inline_ui_html, parse_args
 import asyncio
 import logging
-from starlette.staticfiles import StaticFiles
-import pathlib
-import whisperlivekit.web as webpkg
 
 logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s")
 logging.getLogger().setLevel(logging.WARNING)
@@ -18,16 +15,7 @@ args = parse_args()
 transcription_engine = None
 
 @asynccontextmanager
-async def lifespan(app: FastAPI):
-    
-    #to remove after 0.2.8
-    if args.backend == "simulstreaming" and not args.disable_fast_encoder:
-        logger.warning(f"""
-{'='*50}
-WhisperLiveKit 0.2.8 has introduced a new fast encoder feature using MLX Whisper or Faster Whisper for improved speed. Use --disable-fast-encoder to disable if you encounter issues.
-{'='*50}
-    """)
-    
+async def lifespan(app: FastAPI):    
     global transcription_engine
     transcription_engine = TranscriptionEngine(
         **vars(args),
@@ -42,8 +30,6 @@ app.add_middleware(
     allow_methods=["*"],
     allow_headers=["*"],
 )
-web_dir = pathlib.Path(webpkg.__file__).parent
-app.mount("/web", StaticFiles(directory=str(web_dir)), name="web")
 
 @app.get("/")
 async def get():
@@ -54,7 +40,7 @@ async def handle_websocket_results(websocket, results_generator):
     """Consumes results from the audio processor and sends them via WebSocket."""
     try:
         async for response in results_generator:
-            await websocket.send_json(response)
+            await websocket.send_json(response.to_dict())
         # when the results_generator finishes it means all audio has been processed
         logger.info("Results generator finished. Sending 'ready_to_stop' to client.")
         await websocket.send_json({"type": "ready_to_stop"})
@@ -72,6 +58,11 @@ async def websocket_endpoint(websocket: WebSocket):
     )
     await websocket.accept()
     logger.info("WebSocket connection opened.")
+
+    try:
+        await websocket.send_json({"type": "config", "useAudioWorklet": bool(args.pcm_input)})
+    except Exception as e:
+        logger.warning(f"Failed to send config to client: {e}")
             
     results_generator = await audio_processor.create_tasks()
     websocket_task = asyncio.create_task(handle_websocket_results(websocket, results_generator))
@@ -127,6 +118,8 @@ def main():
 
     if ssl_kwargs:
         uvicorn_kwargs = {**uvicorn_kwargs, **ssl_kwargs}
+    if args.forwarded_allow_ips:
+        uvicorn_kwargs = { **uvicorn_kwargs, "forwarded_allow_ips" : args.forwarded_allow_ips }
 
     uvicorn.run(**uvicorn_kwargs)
 

whisperlivekit/core.py

@@ -4,10 +4,15 @@ try:
 except ImportError:
     from .whisper_streaming_custom.whisper_online import backend_factory
     from .whisper_streaming_custom.online_asr import OnlineASRProcessor
-from whisperlivekit.warmup import warmup_asr, warmup_online
 from argparse import Namespace
 import sys
 
+def update_with_kwargs(_dict, kwargs):
+    _dict.update({
+        k: v for k, v in kwargs.items() if k in _dict
+    })
+    return _dict
+
 class TranscriptionEngine:
     _instance = None
     _initialized = False
@@ -21,69 +26,49 @@ class TranscriptionEngine:
         if TranscriptionEngine._initialized:
             return
 
-        defaults = {
+        global_params = {
             "host": "localhost",
             "port": 8000,
-            "warmup_file": None,
             "diarization": False,
             "punctuation_split": False,
-            "min_chunk_size": 0.5,
-            "model": "tiny",
-            "model_cache_dir": None,
-            "model_dir": None,
-            "lan": "auto",
-            "task": "transcribe",
             "target_language": "",
-            "backend": "faster-whisper",
             "vac": True,
+            "vac_onnx": False,
             "vac_chunk_size": 0.04,
             "log_level": "DEBUG",
             "ssl_certfile": None,
             "ssl_keyfile": None,
+            "forwarded_allow_ips": None,
             "transcription": True,
             "vad": True,
             "pcm_input": False,
-            # whisperstreaming params:
-            "buffer_trimming": "segment",
-            "confidence_validation": False,
-            "buffer_trimming_sec": 15,
-            # simulstreaming params:
-            "disable_fast_encoder": False,
-            "frame_threshold": 25,
-            "beams": 1,
-            "decoder_type": None,
-            "audio_max_len": 20.0,
-            "audio_min_len": 0.0,
-            "cif_ckpt_path": None,
-            "never_fire": False,
-            "init_prompt": None,
-            "static_init_prompt": None,
-            "max_context_tokens": None,
-            "model_path": './base.pt',
-            "diarization_backend": "sortformer",
-            # diarization params:
             "disable_punctuation_split" : False,
-            "segmentation_model": "pyannote/segmentation-3.0",
-            "embedding_model": "pyannote/embedding",         
+            "diarization_backend": "sortformer",
         }
+        global_params = update_with_kwargs(global_params, kwargs)
 
-        config_dict = {**defaults, **kwargs}
+        transcription_common_params = {
+            "backend": "simulstreaming",
+            "warmup_file": None,
+            "min_chunk_size": 0.5,
+            "model_size": "tiny",
+            "model_cache_dir": None,
+            "model_dir": None,
+            "lan": "auto",
+            "task": "transcribe",
+        }
+        transcription_common_params = update_with_kwargs(transcription_common_params, kwargs)                                            
 
+        if transcription_common_params['model_size'].endswith(".en"):
+            transcription_common_params["lan"] = "en"
         if 'no_transcription' in kwargs:
-            config_dict['transcription'] = not kwargs['no_transcription']
+            global_params['transcription'] = not global_params['no_transcription']
         if 'no_vad' in kwargs:
-            config_dict['vad'] = not kwargs['no_vad']
+            global_params['vad'] = not kwargs['no_vad']
         if 'no_vac' in kwargs:
-            config_dict['vac'] = not kwargs['no_vac']
-        
-        config_dict.pop('no_transcription', None)
-        config_dict.pop('no_vad', None)
+            global_params['vac'] = not kwargs['no_vac']
 
-        if 'language' in kwargs:
-            config_dict['lan'] = kwargs['language']
-        config_dict.pop('language', None) 
-
-        self.args = Namespace(**config_dict)
+        self.args = Namespace(**{**global_params, **transcription_common_params})
         
         self.asr = None
         self.tokenizer = None
@@ -91,79 +76,90 @@ class TranscriptionEngine:
         self.vac_model = None
         
         if self.args.vac:
-            import torch
-            self.vac_model, _ = torch.hub.load(repo_or_dir="snakers4/silero-vad", model="silero_vad")            
+            from whisperlivekit.silero_vad_iterator import load_silero_vad
+            # Use ONNX if specified, otherwise use JIT (default)
+            use_onnx = kwargs.get('vac_onnx', False)
+            self.vac_model = load_silero_vad(onnx=use_onnx)
         
         if self.args.transcription:
             if self.args.backend == "simulstreaming": 
                 from whisperlivekit.simul_whisper import SimulStreamingASR
-                self.tokenizer = None
-                simulstreaming_kwargs = {}
-                for attr in ['frame_threshold', 'beams', 'decoder_type', 'audio_max_len', 'audio_min_len', 
-                            'cif_ckpt_path', 'never_fire', 'init_prompt', 'static_init_prompt', 
-                            'max_context_tokens', 'model_path', 'warmup_file', 'preload_model_count', 'disable_fast_encoder']:
-                    if hasattr(self.args, attr):
-                        simulstreaming_kwargs[attr] = getattr(self.args, attr)
-        
-                # Add segment_length from min_chunk_size
-                simulstreaming_kwargs['segment_length'] = getattr(self.args, 'min_chunk_size', 0.5)
-                simulstreaming_kwargs['task'] = self.args.task
                 
-                size = self.args.model
+                simulstreaming_params = {
+                    "disable_fast_encoder": False,
+                    "custom_alignment_heads": None,
+                    "frame_threshold": 25,
+                    "beams": 1,
+                    "decoder_type": None,
+                    "audio_max_len": 20.0,
+                    "audio_min_len": 0.0,
+                    "cif_ckpt_path": None,
+                    "never_fire": False,
+                    "init_prompt": None,
+                    "static_init_prompt": None,
+                    "max_context_tokens": None,
+                    "model_path": './base.pt',
+                    "preload_model_count": 1,
+                }
+                simulstreaming_params = update_with_kwargs(simulstreaming_params, kwargs)
+                
+                self.tokenizer = None        
                 self.asr = SimulStreamingASR(
-                    modelsize=size,
-                    lan=self.args.lan,
-                    cache_dir=getattr(self.args, 'model_cache_dir', None),
-                    model_dir=getattr(self.args, 'model_dir', None),
-                    **simulstreaming_kwargs
+                    **transcription_common_params, **simulstreaming_params
                 )
-
             else:
-                self.asr, self.tokenizer = backend_factory(self.args)
-            warmup_asr(self.asr, self.args.warmup_file) #for simulstreaming, warmup should be done in the online class not here
+                
+                whisperstreaming_params = {
+                    "buffer_trimming": "segment",
+                    "confidence_validation": False,
+                    "buffer_trimming_sec": 15,
+                }
+                whisperstreaming_params = update_with_kwargs(whisperstreaming_params, kwargs)
+                
+                self.asr = backend_factory(
+                    **transcription_common_params, **whisperstreaming_params
+                )
 
         if self.args.diarization:
             if self.args.diarization_backend == "diart":
                 from whisperlivekit.diarization.diart_backend import DiartDiarization
+                diart_params = {
+                    "segmentation_model": "pyannote/segmentation-3.0",
+                    "embedding_model": "pyannote/embedding",
+                }
+                diart_params = update_with_kwargs(diart_params, kwargs)
                 self.diarization_model = DiartDiarization(
                     block_duration=self.args.min_chunk_size,
-                    segmentation_model_name=self.args.segmentation_model,
-                    embedding_model_name=self.args.embedding_model
+                    **diart_params
                 )
             elif self.args.diarization_backend == "sortformer":
                 from whisperlivekit.diarization.sortformer_backend import SortformerDiarization
                 self.diarization_model = SortformerDiarization()
-            else:
-                raise ValueError(f"Unknown diarization backend: {self.args.diarization_backend}")
         
         self.translation_model = None
         if self.args.target_language:
-            if self.args.lan == 'auto':
-                raise Exception('Translation cannot be set with language auto')
+            if self.args.lan == 'auto' and self.args.backend != "simulstreaming":
+                raise Exception('Translation cannot be set with language auto when transcription backend is not simulstreaming')
             else:
-                from whisperlivekit.translation.translation import load_model
-                self.translation_model = load_model([self.args.lan]) #in the future we want to handle different languages for different speakers
-            
+                try:
+                    from nllw import load_model
+                except:
+                    raise Exception('To use translation, you must install nllw: `pip install nllw`')
+                translation_params = { 
+                    "nllb_backend": "transformers",
+                    "nllb_size": "600M"
+                }
+                translation_params = update_with_kwargs(translation_params, kwargs)
+                self.translation_model = load_model([self.args.lan], **translation_params) #in the future we want to handle different languages for different speakers
         TranscriptionEngine._initialized = True
 
 
-
-def online_factory(args, asr, tokenizer, logfile=sys.stderr):
+def online_factory(args, asr):
     if args.backend == "simulstreaming":    
         from whisperlivekit.simul_whisper import SimulStreamingOnlineProcessor
-        online = SimulStreamingOnlineProcessor(
-            asr,
-            logfile=logfile,
-        )
-        # warmup_online(online, args.warmup_file)
+        online = SimulStreamingOnlineProcessor(asr)
     else:
-        online = OnlineASRProcessor(
-            asr,
-            tokenizer,
-            logfile=logfile,
-            buffer_trimming=(args.buffer_trimming, args.buffer_trimming_sec),
-            confidence_validation = args.confidence_validation
-        )
+        online = OnlineASRProcessor(asr)
     return online
   
   
@@ -182,5 +178,6 @@ def online_translation_factory(args, translation_model):
     #should be at speaker level in the future:
     #one shared nllb model for all speaker
     #one tokenizer per speaker/language
-    from whisperlivekit.translation.translation import OnlineTranslation
-    return OnlineTranslation(translation_model, [args.lan], [args.target_language])
+    from nllw import OnlineTranslation
+    from nllw import OnlineTranslation
+    return OnlineTranslation(translation_model, [args.lan], [args.target_language])

whisperlivekit/diarization/sortformer_backend.py

@@ -289,6 +289,7 @@ class SortformerDiarizationOnline:
             
         Returns:
             List of tokens with speaker assignments
+            Last speaker_segment
         """
         with self.segment_lock:
             segments = self.speaker_segments.copy()

whisperlivekit/ffmpeg_manager.py

@@ -7,11 +7,12 @@ import contextlib
 logger = logging.getLogger(__name__)
 logging.basicConfig(level=logging.INFO)
 
-ERROR_INSTALL_INSTRUCTIONS = """
+ERROR_INSTALL_INSTRUCTIONS = f"""
+{'='*50}
 FFmpeg is not installed or not found in your system's PATH.
-Please install FFmpeg to enable audio processing.
+Alternative Solution: You can still use WhisperLiveKit without FFmpeg by adding the --pcm-input parameter. Note that when using this option, audio will not be compressed between the frontend and backend, which may result in higher bandwidth usage.
 
-Installation instructions:
+If you want to install FFmpeg:
 
 # Ubuntu/Debian:
 sudo apt update && sudo apt install ffmpeg
@@ -25,6 +26,7 @@ brew install ffmpeg
 # 3. Add the 'bin' directory (e.g., C:\\FFmpeg\\bin) to your system's PATH environment variable.
 
 After installation, please restart the application.
+{'='*50}
 """
 
 class FFmpegState(Enum):
@@ -183,6 +185,8 @@ class FFmpegManager:
     async def _drain_stderr(self):
         try:
             while True:
+                if not self.process or not self.process.stderr:
+                    break
                 line = await self.process.stderr.readline()
                 if not line:
                     break
@@ -190,4 +194,4 @@ class FFmpegManager:
         except asyncio.CancelledError:
             logger.info("FFmpeg stderr drain task cancelled.")
         except Exception as e:
-            logger.error(f"Error draining FFmpeg stderr: {e}")
+            logger.error(f"Error draining FFmpeg stderr: {e}")

whisperlivekit/parse_args.py

@@ -20,7 +20,7 @@ def parse_args():
         help="""
         The path to a speech audio wav file to warm up Whisper so that the very first chunk processing is fast.
         If not set, uses https://github.com/ggerganov/whisper.cpp/raw/master/samples/jfk.wav.
-        If False, no warmup is performed.
+        If empty, no warmup is performed.
         """,
     )
 
@@ -89,6 +89,7 @@ def parse_args():
         "--model",
         type=str,
         default="small",
+        dest='model_size',
         help="Name size of the Whisper model to use (default: tiny). Suggested values: tiny.en,tiny,base.en,base,small.en,small,medium.en,medium,large-v1,large-v2,large-v3,large,large-v3-turbo. The model is automatically downloaded from the model hub if not present in model cache dir.",
     )
     
@@ -109,6 +110,7 @@ def parse_args():
         "--language",
         type=str,
         default="auto",
+        dest='lan',
         help="Source language code, e.g. en,de,cs, or 'auto' for language detection.",
     )
     parser.add_argument(
@@ -173,11 +175,12 @@ def parse_args():
     )
     parser.add_argument("--ssl-certfile", type=str, help="Path to the SSL certificate file.", default=None)
     parser.add_argument("--ssl-keyfile", type=str, help="Path to the SSL private key file.", default=None)
+    parser.add_argument("--forwarded-allow-ips", type=str, help="Allowed ips for reverse proxying.", default=None)
     parser.add_argument(
         "--pcm-input",
         action="store_true",
         default=False,
-        help="If set, raw PCM (s16le) data is expected as input and FFmpeg will be bypassed."
+        help="If set, raw PCM (s16le) data is expected as input and FFmpeg will be bypassed. Frontend will use AudioWorklet instead of MediaRecorder."
     )
     # SimulStreaming-specific arguments
     simulstreaming_group = parser.add_argument_group('SimulStreaming arguments (only used with --backend simulstreaming)')
@@ -189,6 +192,13 @@ def parse_args():
         dest="disable_fast_encoder",
         help="Disable Faster Whisper or MLX Whisper backends for encoding (if installed). Slower but helpful when GPU memory is limited",
     )
+
+    simulstreaming_group.add_argument(
+        "--custom-alignment-heads",
+        type=str,
+        default=None,
+        help="Use your own alignment heads, useful when `--model-dir` is used",
+    )
     
     simulstreaming_group.add_argument(
         "--frame-threshold",
@@ -287,6 +297,20 @@ def parse_args():
         help="Optional. Number of models to preload in memory to speed up loading (set up to the expected number of concurrent instances).",
     )
 
+    simulstreaming_group.add_argument(
+        "--nllb-backend",
+        type=str,
+        default="transformers",
+        help="transformers or ctranslate2",
+    )
+    
+    simulstreaming_group.add_argument(
+        "--nllb-size",
+        type=str,
+        default="600M",
+        help="600M or 1.3B",
+    )
+
     args = parser.parse_args()
     
     args.transcription = not args.no_transcription

whisperlivekit/remove_silences.py

@@ -1,4 +1,5 @@
 from whisperlivekit.timed_objects import ASRToken
+from time import time
 import re
 
 MIN_SILENCE_DURATION = 4 #in seconds
@@ -39,7 +40,7 @@ def blank_to_silence(tokens):
                         )
                 else:
                     if silence_token: #there was silence but no more
-                        if silence_token.end - silence_token.start >= MIN_SILENCE_DURATION:
+                        if silence_token.duration() >= MIN_SILENCE_DURATION:
                             cleaned_tokens.append(
                                 silence_token
                             )
@@ -77,15 +78,11 @@ def no_token_to_silence(tokens):
             new_tokens.append(token)
     return new_tokens
             
-def ends_with_silence(tokens, buffer_transcription, buffer_diarization, current_time, vac_detected_silence):
-    if not tokens:
-        return [], buffer_transcription, buffer_diarization
+def ends_with_silence(tokens, beg_loop, vac_detected_silence):
+    current_time = time() - (beg_loop if beg_loop else 0.0)
     last_token = tokens[-1]
-    if tokens and current_time and (
-        current_time - last_token.end >= END_SILENCE_DURATION 
-        or 
-        (current_time - last_token.end >= 3 and vac_detected_silence)
-        ):
+    silence_duration = current_time - last_token.end
+    if  (vac_detected_silence and silence_duration > END_SILENCE_DURATION_VAC) or (silence_duration >= END_SILENCE_DURATION):
         if last_token.speaker == -2:
             last_token.end = current_time
         else:
@@ -97,14 +94,14 @@ def ends_with_silence(tokens, buffer_transcription, buffer_diarization, current_
                     probability=0.95
                 )
             )
-        buffer_transcription = "" # for whisperstreaming backend, we should probably validate the buffer has because of the silence
-        buffer_diarization  = ""
-    return tokens, buffer_transcription, buffer_diarization
+    return tokens
     
 
-def handle_silences(tokens, buffer_transcription, buffer_diarization, current_time, vac_detected_silence):
+def handle_silences(tokens, beg_loop, vac_detected_silence):
+    if not tokens:
+        return []
     tokens = blank_to_silence(tokens) #useful for simulstreaming backend which tends to generate [BLANK_AUDIO] text
     tokens = no_token_to_silence(tokens)
-    tokens, buffer_transcription, buffer_diarization = ends_with_silence(tokens, buffer_transcription, buffer_diarization, current_time, vac_detected_silence)
-    return tokens, buffer_transcription, buffer_diarization
+    tokens = ends_with_silence(tokens, beg_loop, vac_detected_silence)
+    return tokens
      

whisperlivekit/results_formater.py

@@ -1,16 +1,17 @@
 
 import logging
 from whisperlivekit.remove_silences import handle_silences
-from whisperlivekit.timed_objects import Line, format_time
+from whisperlivekit.timed_objects import Line, Segment, format_time
 
 logger = logging.getLogger(__name__)
 logger.setLevel(logging.DEBUG)
 
-PUNCTUATION_MARKS = {'.', '!', '?', '。', '！', '？'}
 CHECK_AROUND = 4
+DEBUG = False
+
 
 def is_punctuation(token):
-    if token.text.strip() in PUNCTUATION_MARKS:
+    if token.is_punctuation():
         return True
     return False
 
@@ -31,107 +32,136 @@ def next_speaker_change(i, tokens, speaker):
     
 def new_line(
     token,
-    speaker,
-    debug_info = ""
 ):
     return Line(
-        speaker = speaker,
-        text = token.text + debug_info,
+        speaker = token.corrected_speaker,
+        text = token.text + (f"[{format_time(token.start)} : {format_time(token.end)}]" if DEBUG else ""),
         start = token.start,
         end = token.end,
+        detected_language=token.detected_language
     )
 
-def append_token_to_last_line(lines, sep, token, debug_info):
-    if token.text:
-        lines[-1].text += sep + token.text + debug_info
-        lines[-1].end = token.end
+def append_token_to_last_line(lines, sep, token):
+    if not lines:
+        lines.append(new_line(token))
+    else:
+        if token.text:
+            lines[-1].text += sep + token.text + (f"[{format_time(token.start)} : {format_time(token.end)}]" if DEBUG else "")
+            lines[-1].end = token.end
+        if not lines[-1].detected_language and token.detected_language:
+            lines[-1].detected_language = token.detected_language
+            
 
-def format_output(state, silence, current_time, args, debug):
+def format_output(state, silence, args, sep):
     diarization = args.diarization
     disable_punctuation_split = args.disable_punctuation_split
-    tokens = state["tokens"]
-    translated_segments = state["translated_segments"] # Here we will attribute the speakers only based on the timestamps of the segments
-    buffer_transcription = state["buffer_transcription"]
-    buffer_diarization = state["buffer_diarization"]
-    end_attributed_speaker = state["end_attributed_speaker"]
-    sep = state["sep"]
+    tokens = state.tokens
+    translation_validated_segments = state.translation_validated_segments # Here we will attribute the speakers only based on the timestamps of the segments
+    translation_buffer = state.translation_buffer
+    last_validated_token = state.last_validated_token
     
-    previous_speaker = -1
-    lines = []
+    previous_speaker = 1
     undiarized_text = []
-    tokens, buffer_transcription, buffer_diarization = handle_silences(tokens, buffer_transcription, buffer_diarization, current_time, silence)
+    tokens = handle_silences(tokens, state.beg_loop, silence)
     last_punctuation = None
-    for i, token in enumerate(tokens):
-        speaker = token.speaker
-        if not diarization and speaker == -1: #Speaker -1 means no attributed by diarization. In the frontend, it should appear under 'Speaker 1'
-            speaker = 1
-        if diarization and not tokens[-1].speaker == -2:
-            if (speaker in [-1, 0]) and token.end >= end_attributed_speaker:
-                undiarized_text.append(token.text)
-                continue
-            elif (speaker in [-1, 0]) and token.end < end_attributed_speaker:
-                speaker = previous_speaker
-        debug_info = ""
-        if debug:
-            debug_info = f"[{format_time(token.start)} : {format_time(token.end)}]"
-            
-        if not lines:
-            lines.append(new_line(token, speaker, debug_info = ""))
-            continue
+    for i, token in enumerate(tokens[last_validated_token:]):
+        speaker = int(token.speaker)
+        token.corrected_speaker = speaker
+        if not diarization:
+            if speaker == -1: #Speaker -1 means no attributed by diarization. In the frontend, it should appear under 'Speaker 1'
+                token.corrected_speaker = 1
+                token.validated_speaker = True
         else:
-            previous_speaker = lines[-1].speaker
-        
-        if is_punctuation(token):
-            last_punctuation = i
+            if is_punctuation(token):
+                last_punctuation = i  
             
-        
-        if last_punctuation == i-1:
-            if speaker != previous_speaker:
-                # perfect, diarization perfectly aligned
-                lines.append(new_line(token, speaker, debug_info = ""))
-                last_punctuation, next_punctuation = None, None
-                continue
-            
-            speaker_change_pos, new_speaker = next_speaker_change(i, tokens, speaker)
-            if speaker_change_pos:
-                # Corrects delay:
-                # That was the idea. Okay haha |SPLIT SPEAKER| that's a good one 
-                # should become:
-                # That was the idea. |SPLIT SPEAKER| Okay haha that's a good one 
-                lines.append(new_line(token, new_speaker, debug_info = ""))
-            else:
-                # No speaker change to come
-                append_token_to_last_line(lines, sep, token, debug_info)
-            continue
-        
-
-        if speaker != previous_speaker:
-            if speaker == -2 or previous_speaker == -2: #silences can happen anytime
-                lines.append(new_line(token, speaker, debug_info = ""))
-                continue
-            elif next_punctuation_change(i, tokens):
-                # Corrects advance:
-                # Are you |SPLIT SPEAKER| okay? yeah, sure. Absolutely 
-                # should become:
-                # Are you okay? |SPLIT SPEAKER| yeah, sure. Absolutely 
-                append_token_to_last_line(lines, sep, token, debug_info)
-                continue
-            else: #we create a new speaker, but that's no ideal. We are not sure about the split. We prefer to append to previous line
-                if disable_punctuation_split:
-                    lines.append(new_line(token, speaker, debug_info = ""))
-                    continue
-                pass
-            
-        append_token_to_last_line(lines, sep, token, debug_info)
-    if lines and translated_segments:
-        cts_idx = 0 # current_translated_segment_idx
-        for line in lines:
-            while cts_idx < len(translated_segments):
-                ts = translated_segments[cts_idx]
-                if ts.start and ts.start >= line.start and ts.end <= line.end:
-                    line.translation += ts.text + ' '
-                    cts_idx += 1
+            if last_punctuation == i-1:
+                if token.speaker != previous_speaker:
+                    token.validated_speaker = True
+                    # perfect, diarization perfectly aligned
+                    last_punctuation = None
                 else:
-                    break
-    return lines, undiarized_text, buffer_transcription, '' 
+                    speaker_change_pos, new_speaker = next_speaker_change(i, tokens, speaker)
+                    if speaker_change_pos:
+                        # Corrects delay:
+                        # That was the idea. <Okay> haha |SPLIT SPEAKER| that's a good one 
+                        # should become:
+                        # That was the idea. |SPLIT SPEAKER| <Okay> haha that's a good one 
+                        token.corrected_speaker = new_speaker
+                        token.validated_speaker = True
+            elif speaker != previous_speaker:
+                if not (speaker == -2 or previous_speaker == -2):
+                    if next_punctuation_change(i, tokens):
+                        # Corrects advance:
+                        # Are you |SPLIT SPEAKER| <okay>? yeah, sure. Absolutely 
+                        # should become:
+                        # Are you <okay>? |SPLIT SPEAKER| yeah, sure. Absolutely 
+                        token.corrected_speaker = previous_speaker
+                        token.validated_speaker = True
+                    else: #Problematic, except if the language has no punctuation. We append to previous line, except if disable_punctuation_split is set to True.
+                        if not disable_punctuation_split:
+                            token.corrected_speaker = previous_speaker
+                            token.validated_speaker = False
+        if token.validated_speaker:
+            state.last_validated_token = i + last_validated_token
+        previous_speaker = token.corrected_speaker  
 
+    for token in tokens[last_validated_token+1:state.last_validated_token+1]:
+        if not state.segments or int(token.corrected_speaker) != int(state.segments[-1].speaker):
+            state.segments.append(
+                Segment(
+                    speaker=token.corrected_speaker,
+                    words=[token]
+                )
+            )
+        else:
+            state.segments[-1].words.append(token)
+ 
+    for token in tokens[state.last_validated_token+1:]:
+        # if not state.segments or int(token.corrected_speaker) != int(state.segments[-1].speaker):
+        #     state.segments.append(
+        #         Segment(
+        #             speaker=token.corrected_speaker,
+        #             buffer_tokens=[token]
+        #         )
+        #     )
+        # else:
+        state.segments[-1].buffer_tokens.append(token) 
+    
+    for segment in state.segments:
+        segment.consolidate(sep)
+    # lines = []
+    # for token in tokens:
+    #     if int(token.corrected_speaker) != int(previous_speaker):
+    #         lines.append(new_line(token))
+    #     else:
+    #         append_token_to_last_line(lines, sep, token)
+
+    #     previous_speaker = token.corrected_speaker
+
+    for ts in translation_validated_segments:
+        for segment in state.segments[state.last_validated_segment:]:
+            if ts.is_within(segment):
+                segment.translation += ts.text + sep
+                break
+
+    for ts in translation_buffer:
+        for segment in state.segments[state.last_validated_segment:]:
+            if ts.is_within(segment):
+                segment.buffer.translation += ts.text + sep
+                break
+    
+    # if state.buffer_transcription and lines:
+    #     lines[-1].end = max(state.buffer_transcription.end, lines[-1].end)
+    
+    lines = []
+    for segment in state.segments:
+        lines.append(Line(
+            start=segment.start,
+            end=segment.end,
+            speaker=segment.speaker,
+            text=segment.text,
+            translation=segment.translation
+        ))
+    
+    return lines, undiarized_text

whisperlivekit/silero_vad_iterator.py

@@ -1,27 +1,182 @@
 import torch
+import numpy as np
+import warnings
+from pathlib import Path
 
-# This is copied from silero-vad's vad_utils.py:
-# https://github.com/snakers4/silero-vad/blob/f6b1294cb27590fb2452899df98fb234dfef1134/utils_vad.py#L340
-# (except changed defaults)
+"""
+Code is adapted from silero-vad v6: https://github.com/snakers4/silero-vad
+"""
 
-# Their licence is MIT, same as ours: https://github.com/snakers4/silero-vad/blob/f6b1294cb27590fb2452899df98fb234dfef1134/LICENSE
+def init_jit_model(model_path: str, device=torch.device('cpu')):
+    """Load a JIT model from file."""
+    model = torch.jit.load(model_path, map_location=device)
+    model.eval()
+    return model
+
+
+class OnnxWrapper():
+    """ONNX Runtime wrapper for Silero VAD model."""
+
+    def __init__(self, path, force_onnx_cpu=False):
+        global np
+        import numpy as np
+        import onnxruntime
+
+        opts = onnxruntime.SessionOptions()
+        opts.inter_op_num_threads = 1
+        opts.intra_op_num_threads = 1
+
+        if force_onnx_cpu and 'CPUExecutionProvider' in onnxruntime.get_available_providers():
+            self.session = onnxruntime.InferenceSession(path, providers=['CPUExecutionProvider'], sess_options=opts)
+        else:
+            self.session = onnxruntime.InferenceSession(path, sess_options=opts)
+
+        self.reset_states()
+        if '16k' in path:
+            warnings.warn('This model support only 16000 sampling rate!')
+            self.sample_rates = [16000]
+        else:
+            self.sample_rates = [8000, 16000]
+
+    def _validate_input(self, x, sr: int):
+        if x.dim() == 1:
+            x = x.unsqueeze(0)
+        if x.dim() > 2:
+            raise ValueError(f"Too many dimensions for input audio chunk {x.dim()}")
+
+        if sr != 16000 and (sr % 16000 == 0):
+            step = sr // 16000
+            x = x[:,::step]
+            sr = 16000
+
+        if sr not in self.sample_rates:
+            raise ValueError(f"Supported sampling rates: {self.sample_rates} (or multiply of 16000)")
+        if sr / x.shape[1] > 31.25:
+            raise ValueError("Input audio chunk is too short")
+
+        return x, sr
+
+    def reset_states(self, batch_size=1):
+        self._state = torch.zeros((2, batch_size, 128)).float()
+        self._context = torch.zeros(0)
+        self._last_sr = 0
+        self._last_batch_size = 0
+
+    def __call__(self, x, sr: int):
+
+        x, sr = self._validate_input(x, sr)
+        num_samples = 512 if sr == 16000 else 256
+
+        if x.shape[-1] != num_samples:
+            raise ValueError(f"Provided number of samples is {x.shape[-1]} (Supported values: 256 for 8000 sample rate, 512 for 16000)")
+
+        batch_size = x.shape[0]
+        context_size = 64 if sr == 16000 else 32
+
+        if not self._last_batch_size:
+            self.reset_states(batch_size)
+        if (self._last_sr) and (self._last_sr != sr):
+            self.reset_states(batch_size)
+        if (self._last_batch_size) and (self._last_batch_size != batch_size):
+            self.reset_states(batch_size)
+
+        if not len(self._context):
+            self._context = torch.zeros(batch_size, context_size)
+
+        x = torch.cat([self._context, x], dim=1)
+        if sr in [8000, 16000]:
+            ort_inputs = {'input': x.numpy(), 'state': self._state.numpy(), 'sr': np.array(sr, dtype='int64')}
+            ort_outs = self.session.run(None, ort_inputs)
+            out, state = ort_outs
+            self._state = torch.from_numpy(state)
+        else:
+            raise ValueError()
+
+        self._context = x[..., -context_size:]
+        self._last_sr = sr
+        self._last_batch_size = batch_size
+
+        out = torch.from_numpy(out)
+        return out
+
+
+def load_silero_vad(model_path: str = None, onnx: bool = False, opset_version: int = 16):
+    """
+    Load Silero VAD model (JIT or ONNX).
+    
+    Parameters
+    ----------
+    model_path : str, optional
+        Path to model file. If None, uses default bundled model.
+    onnx : bool, default False
+        Whether to use ONNX runtime (requires onnxruntime package).
+    opset_version : int, default 16
+        ONNX opset version (15 or 16). Only used if onnx=True.
+    
+    Returns
+    -------
+    model
+        Loaded VAD model (JIT or ONNX wrapper)
+    """
+    available_ops = [15, 16]
+    if onnx and opset_version not in available_ops:
+        raise Exception(f'Available ONNX opset_version: {available_ops}')
+    if model_path is None:
+        current_dir = Path(__file__).parent
+        data_dir = current_dir / 'vad_models'
+        
+        if onnx:
+            if opset_version == 16:
+                model_name = 'silero_vad.onnx'
+            else:
+                model_name = f'silero_vad_16k_op{opset_version}.onnx'
+        else:
+            model_name = 'silero_vad.jit'
+        
+        model_path = data_dir / model_name
+        
+        if not model_path.exists():
+            raise FileNotFoundError(
+                f"Model file not found: {model_path}\n"
+                f"Please ensure the whisperlivekit/vad_models/ directory contains the model files."
+            )
+    else:
+        model_path = Path(model_path)
+    if onnx:
+        try:
+            model = OnnxWrapper(str(model_path), force_onnx_cpu=True)
+        except ImportError:
+            raise ImportError(
+                "ONNX runtime not available. Install with: pip install onnxruntime\n"
+                "Or use JIT model by setting onnx=False"
+            )
+    else:
+        model = init_jit_model(str(model_path))
+
+    return model
 
 
 class VADIterator:
-    def __init__(
-        self,
-        model,
-        threshold: float = 0.5,
-        sampling_rate: int = 16000,
-        min_silence_duration_ms: int = 500,  # makes sense on one recording that I checked
-        speech_pad_ms: int = 100,  # same
-    ):
+    """
+    Voice Activity Detection iterator for streaming audio.
+    
+    This is the Silero VAD v6 implementation.
+    """
+    
+    def __init__(self,
+                 model,
+                 threshold: float = 0.5,
+                 sampling_rate: int = 16000,
+                 min_silence_duration_ms: int = 100,
+                 speech_pad_ms: int = 30
+                 ):
+
         """
         Class for stream imitation
 
         Parameters
         ----------
-        model: preloaded .jit silero VAD model
+        model: preloaded .jit/.onnx silero VAD model
 
         threshold: float (default - 0.5)
             Speech threshold. Silero VAD outputs speech probabilities for each audio chunk, probabilities ABOVE this value are considered as SPEECH.
@@ -42,9 +197,7 @@ class VADIterator:
         self.sampling_rate = sampling_rate
 
         if sampling_rate not in [8000, 16000]:
-            raise ValueError(
-                "VADIterator does not support sampling rates other than [8000, 16000]"
-            )
+            raise ValueError('VADIterator does not support sampling rates other than [8000, 16000]')
 
         self.min_silence_samples = sampling_rate * min_silence_duration_ms / 1000
         self.speech_pad_samples = sampling_rate * speech_pad_ms / 1000
@@ -57,13 +210,17 @@ class VADIterator:
         self.temp_end = 0
         self.current_sample = 0
 
-    def __call__(self, x, return_seconds=False):
+    @torch.no_grad()
+    def __call__(self, x, return_seconds=False, time_resolution: int = 1):
         """
         x: torch.Tensor
             audio chunk (see examples in repo)
 
         return_seconds: bool (default - False)
             whether return timestamps in seconds (default - samples)
+
+        time_resolution: int (default - 1)
+            time resolution of speech coordinates when requested as seconds
         """
 
         if not torch.is_tensor(x):
@@ -82,14 +239,8 @@ class VADIterator:
 
         if (speech_prob >= self.threshold) and not self.triggered:
             self.triggered = True
-            speech_start = self.current_sample - self.speech_pad_samples
-            return {
-                "start": (
-                    int(speech_start)
-                    if not return_seconds
-                    else round(speech_start / self.sampling_rate, 1)
-                )
-            }
+            speech_start = max(0, self.current_sample - self.speech_pad_samples - window_size_samples)
+            return {'start': int(speech_start) if not return_seconds else round(speech_start / self.sampling_rate, time_resolution)}
 
         if (speech_prob < self.threshold - 0.15) and self.triggered:
             if not self.temp_end:
@@ -97,30 +248,17 @@ class VADIterator:
             if self.current_sample - self.temp_end < self.min_silence_samples:
                 return None
             else:
-                speech_end = self.temp_end + self.speech_pad_samples
+                speech_end = self.temp_end + self.speech_pad_samples - window_size_samples
                 self.temp_end = 0
                 self.triggered = False
-                return {
-                    "end": (
-                        int(speech_end)
-                        if not return_seconds
-                        else round(speech_end / self.sampling_rate, 1)
-                    )
-                }
+                return {'end': int(speech_end) if not return_seconds else round(speech_end / self.sampling_rate, time_resolution)}
 
         return None
 
 
-#######################
-# because Silero now requires exactly 512-sized audio chunks
-
-import numpy as np
-
-
 class FixedVADIterator(VADIterator):
-    """It fixes VADIterator by allowing to process any audio length, not only exactly 512 frames at once.
-    If audio to be processed at once is long and multiple voiced segments detected,
-    then __call__ returns the start of the first segment, and end (or middle, which means no end) of the last segment.
+    """
+    Fixed VAD Iterator that handles variable-length audio chunks, not only exactly 512 frames at once.
     """
 
     def reset_states(self):
@@ -137,27 +275,20 @@ class FixedVADIterator(VADIterator):
                 ret = r
             elif r is not None:
                 if "end" in r:
-                    ret["end"] = r["end"]  # the latter end
-                if "start" in r and "end" in ret:  # there is an earlier start.
-                    # Remove end, merging this segment with the previous one.
+                    ret["end"] = r["end"]
+                if "start" in r and "end" in ret:
                     del ret["end"]
         return ret if ret != {} else None
 
 
 if __name__ == "__main__":
-    # test/demonstrate the need for FixedVADIterator:
-
-    import torch
-
-    model, _ = torch.hub.load(repo_or_dir="snakers4/silero-vad", model="silero_vad")
-    vac = FixedVADIterator(model)
-    #   vac = VADIterator(model)  # the second case crashes with this
-
-    # this works: for both
-    audio_buffer = np.array([0] * (512), dtype=np.float32)
-    vac(audio_buffer)
-
-    # this crashes on the non FixedVADIterator with
-    # ops.prim.RaiseException("Input audio chunk is too short", "builtins.ValueError")
-    audio_buffer = np.array([0] * (512 - 1), dtype=np.float32)
-    vac(audio_buffer)
+    model = load_silero_vad(onnx=False)
+    vad = FixedVADIterator(model)
+    
+    audio_buffer = np.array([0] * 512, dtype=np.float32)
+    result = vad(audio_buffer)
+    print(f"   512 samples: {result}")
+    
+    # test with 511 samples
+    audio_buffer = np.array([0] * 511, dtype=np.float32)
+    result = vad(audio_buffer)

whisperlivekit/simul_whisper/backend.py

@@ -4,13 +4,13 @@ import logging
 from typing import List, Tuple, Optional
 import logging
 import platform
-from whisperlivekit.timed_objects import ASRToken, Transcript
+from whisperlivekit.timed_objects import ASRToken, Transcript, ChangeSpeaker
 from whisperlivekit.warmup import load_file
-from whisperlivekit.simul_whisper.license_simulstreaming import SIMULSTREAMING_LICENSE
 from .whisper import load_model, tokenizer
 from .whisper.audio import TOKENS_PER_SECOND
 import os
 import gc
+from pathlib import Path
 logger = logging.getLogger(__name__)
 
 import torch
@@ -23,7 +23,7 @@ try:
     HAS_MLX_WHISPER = True
 except ImportError:
     if platform.system() == "Darwin" and platform.machine() == "arm64":
-        print('MLX Whisper not found but you are on Apple Silicon. Consider installing mlx-whisper for better performance: pip install mlx-whisper')
+        print(f"""{"="*50}\nMLX Whisper not found but you are on Apple Silicon. Consider installing mlx-whisper for better performance: pip install mlx-whisper\n{"="*50}""")
     HAS_MLX_WHISPER = False
 if HAS_MLX_WHISPER:
     HAS_FASTER_WHISPER = False
@@ -34,8 +34,24 @@ else:
     except ImportError:
         HAS_FASTER_WHISPER = False
 
+def model_path_and_type(model_path):
+    path = Path(model_path)
+    
+    compatible_whisper_mlx = False
+    compatible_faster_whisper = False
+    pt_path = path if path.is_file() and path.suffix.lower() == '.pt' else None
+    
+    if path.is_dir():
+        for file in path.iterdir():
+            if file.is_file():
+                if file.name in ['weights.npz', "weights.safetensors"]:
+                    compatible_whisper_mlx = True
+                elif file.suffix.lower() == '.bin':
+                    compatible_faster_whisper = True
+                elif file.suffix.lower() == '.pt':
+                    pt_path = file
+    return pt_path, compatible_whisper_mlx, compatible_faster_whisper
 
-# TOO_MANY_REPETITIONS = 3
 
 class SimulStreamingOnlineProcessor:
     SAMPLING_RATE = 16000
@@ -44,13 +60,11 @@ class SimulStreamingOnlineProcessor:
         self,
         asr,
         logfile=sys.stderr,
-        warmup_file=None
     ):        
         self.asr = asr
         self.logfile = logfile
         self.end = 0.0
-        self.global_time_offset = 0.0
-        
+        self.buffer = []
         self.committed: List[ASRToken] = []
         self.last_result_tokens: List[ASRToken] = []
         self.load_new_backend()
@@ -79,7 +93,7 @@ class SimulStreamingOnlineProcessor:
         else:
             self.process_iter(is_last=True) #we want to totally process what remains in the buffer.
             self.model.refresh_segment(complete=True)
-            self.global_time_offset = silence_duration + offset
+            self.model.global_time_offset = silence_duration + offset
 
 
         
@@ -91,63 +105,15 @@ class SimulStreamingOnlineProcessor:
         self.end = audio_stream_end_time #Only to be aligned with what happens in whisperstreaming backend.
         self.model.insert_audio(audio_tensor)
 
-    def get_buffer(self):
-        return Transcript(
-            start=None, 
-            end=None, 
-            text='', 
-            probability=None
-        )
-
-    def timestamped_text(self, tokens, generation):
-        """
-        generate timestamped text from tokens and generation data.
-        
-        args:
-            tokens: List of tokens to process
-            generation: Dictionary containing generation progress and optionally results
+    def new_speaker(self, change_speaker: ChangeSpeaker):
+            self.process_iter(is_last=True)
+            self.model.refresh_segment(complete=True)
+            self.model.speaker = change_speaker.speaker
+            self.global_time_offset = change_speaker.start
             
-        returns:
-            List of tuples containing (start_time, end_time, word) for each word
-        """
-        FRAME_DURATION = 0.02    
-        if "result" in generation:
-            split_words = generation["result"]["split_words"]
-            split_tokens = generation["result"]["split_tokens"]
-        else:
-            split_words, split_tokens = self.model.tokenizer.split_to_word_tokens(tokens)
-        progress = generation["progress"]
-        frames = [p["most_attended_frames"][0] for p in progress]
-        absolute_timestamps = [p["absolute_timestamps"][0] for p in progress]
-        tokens_queue = tokens.copy()
-        timestamped_words = []
-        
-        for word, word_tokens in zip(split_words, split_tokens):
-            # start_frame = None
-            # end_frame = None
-            for expected_token in word_tokens:
-                if not tokens_queue or not frames:
-                    raise ValueError(f"Insufficient tokens or frames for word '{word}'")
-                    
-                actual_token = tokens_queue.pop(0)
-                current_frame = frames.pop(0)
-                current_timestamp = absolute_timestamps.pop(0)
-                if actual_token != expected_token:
-                    raise ValueError(
-                        f"Token mismatch: expected '{expected_token}', "
-                        f"got '{actual_token}' at frame {current_frame}"
-                    )
-                # if start_frame is None:
-                #     start_frame = current_frame
-                # end_frame = current_frame
-            # start_time = start_frame * FRAME_DURATION
-            # end_time = end_frame * FRAME_DURATION
-            start_time = current_timestamp
-            end_time = current_timestamp + 0.1
-            timestamp_entry = (start_time, end_time, word)
-            timestamped_words.append(timestamp_entry)
-            logger.debug(f"TS-WORD:\t{start_time:.2f}\t{end_time:.2f}\t{word}")
-        return timestamped_words
+    def get_buffer(self):
+        concat_buffer = Transcript.from_tokens(tokens= self.buffer, sep='')
+        return concat_buffer
 
     def process_iter(self, is_last=False) -> Tuple[List[ASRToken], float]:
         """
@@ -156,47 +122,14 @@ class SimulStreamingOnlineProcessor:
         Returns a tuple: (list of committed ASRToken objects, float representing the audio processed up to time).
         """
         try:
-            tokens, generation_progress = self.model.infer(is_last=is_last)
-            ts_words = self.timestamped_text(tokens, generation_progress)
+            timestamped_words = self.model.infer(is_last=is_last)
+            if self.model.cfg.language == "auto" and timestamped_words and timestamped_words[0].detected_language == None:
+                self.buffer.extend(timestamped_words)
+                return [], self.end
             
-            new_tokens = []
-            for ts_word in ts_words:
-                
-                start, end, word = ts_word
-                token = ASRToken(
-                    start=start,
-                    end=end,
-                    text=word,
-                    probability=0.95  # fake prob. Maybe we can extract it from the model?
-                ).with_offset(
-                    self.global_time_offset
-                )
-                new_tokens.append(token)
-                
-            # identical_tokens = 0
-            # n_new_tokens = len(new_tokens)
-            # if n_new_tokens:
-            
-            self.committed.extend(new_tokens)
-            
-            # if token in self.committed:
-            #     pos = len(self.committed) - 1 - self.committed[::-1].index(token)
-            # if pos:
-            #     for i in range(len(self.committed) - n_new_tokens, -1, -n_new_tokens):
-            #         commited_segment = self.committed[i:i+n_new_tokens]
-            #         if commited_segment == new_tokens:
-            #             identical_segments +=1
-            #             if identical_tokens >= TOO_MANY_REPETITIONS:
-            #                 logger.warning('Too many repetition, model is stuck. Load a new one')
-            #                 self.committed = self.committed[:i]
-            #                 self.load_new_backend()
-            #                 return [], self.end
-
-            # pos = self.committed.rindex(token)
-
-            
-            
-            return new_tokens, self.end
+            self.committed.extend(timestamped_words)
+            self.buffer = []
+            return timestamped_words, self.end
 
             
         except Exception as e:
@@ -225,32 +158,23 @@ class SimulStreamingASR():
     """SimulStreaming backend with AlignAtt policy."""
     sep = ""
 
-    def __init__(self, lan, modelsize=None, cache_dir=None, model_dir=None, logfile=sys.stderr, **kwargs):
-        logger.warning(SIMULSTREAMING_LICENSE)
+    def __init__(self, logfile=sys.stderr, **kwargs):
         self.logfile = logfile
         self.transcribe_kargs = {}
-        self.original_language = lan
         
-        self.model_path = kwargs.get('model_path', './large-v3.pt')
-        self.frame_threshold = kwargs.get('frame_threshold', 25)
-        self.audio_max_len = kwargs.get('audio_max_len', 20.0)
-        self.audio_min_len = kwargs.get('audio_min_len', 0.0)
-        self.segment_length = kwargs.get('segment_length', 0.5)
-        self.beams = kwargs.get('beams', 1)
-        self.decoder_type = kwargs.get('decoder_type', 'greedy' if self.beams == 1 else 'beam')
-        self.task = kwargs.get('task', 'transcribe')
-        self.cif_ckpt_path = kwargs.get('cif_ckpt_path', None)
-        self.never_fire = kwargs.get('never_fire', False)
-        self.init_prompt = kwargs.get('init_prompt', None)
-        self.static_init_prompt = kwargs.get('static_init_prompt', None)
-        self.max_context_tokens = kwargs.get('max_context_tokens', None)
-        self.warmup_file = kwargs.get('warmup_file', None)
-        self.preload_model_count = kwargs.get('preload_model_count', 1)
-        self.disable_fast_encoder = kwargs.get('disable_fast_encoder', False)
+        for key, value in kwargs.items():
+            setattr(self, key, value)
+
+        if self.decoder_type is None:
+            self.decoder_type = 'greedy' if self.beams == 1 else 'beam'
+
         self.fast_encoder = False
-        if model_dir is not None:
-            self.model_path = model_dir
-        elif modelsize is not None:
+        
+        pt_path, compatible_whisper_mlx, compatible_faster_whisper = None, True, True
+        if self.model_path:
+            pt_path, compatible_whisper_mlx, compatible_faster_whisper = model_path_and_type(self.model_path)
+            
+        elif self.model_size is not None:
             model_mapping = {
                 'tiny': './tiny.pt',
                 'base': './base.pt',
@@ -265,13 +189,15 @@ class SimulStreamingASR():
                 'large-v3': './large-v3.pt',
                 'large': './large-v3.pt'
             }
-            self.model_path = model_mapping.get(modelsize, f'./{modelsize}.pt')
+            pt_path = Path(model_mapping.get(self.model_size, f'./{self.model_size}.pt'))
+        
+        self.model_name = pt_path.name.replace(".pt", "")
         
         self.cfg = AlignAttConfig(
-                model_path=self.model_path,
-                segment_length=self.segment_length,
+                tokenizer_is_multilingual= not self.model_name.endswith(".en"),
+                segment_length=self.min_chunk_size,
                 frame_threshold=self.frame_threshold,
-                language=self.original_language,
+                language=self.lan,
                 audio_max_len=self.audio_max_len,
                 audio_min_len=self.audio_min_len,
                 cif_ckpt_path=self.cif_ckpt_path,
@@ -290,20 +216,27 @@ class SimulStreamingASR():
         else:
             self.tokenizer = None
         
-        self.model_name = os.path.basename(self.cfg.model_path).replace(".pt", "")
-        self.model_path = os.path.dirname(os.path.abspath(self.cfg.model_path))
+        
+            
     
         self.mlx_encoder, self.fw_encoder = None, None
         if not self.disable_fast_encoder:
             if HAS_MLX_WHISPER:
-                print('Simulstreaming will use MLX whisper for a faster encoder.')
-                mlx_model_name = mlx_model_mapping[self.model_name]
-                self.mlx_encoder = load_mlx_encoder(path_or_hf_repo=mlx_model_name)
+                print('Simulstreaming will use MLX whisper to increase encoding speed.')
+                if self.model_path and compatible_whisper_mlx:
+                    mlx_model = self.model_path
+                else:
+                    mlx_model = mlx_model_mapping[self.model_name]
+                self.mlx_encoder = load_mlx_encoder(path_or_hf_repo=mlx_model)
                 self.fast_encoder = True
-            elif HAS_FASTER_WHISPER:
+            elif HAS_FASTER_WHISPER and compatible_faster_whisper:
                 print('Simulstreaming will use Faster Whisper for the encoder.')
+                if self.model_path and compatible_faster_whisper:
+                    fw_model = self.model_path
+                else:
+                    fw_model = self.model_name
                 self.fw_encoder = WhisperModel(
-                    self.model_name,
+                    fw_model,
                     device='auto',
                     compute_type='auto',
                 )
@@ -313,7 +246,12 @@ class SimulStreamingASR():
 
 
     def load_model(self):
-        whisper_model = load_model(name=self.model_name, download_root=self.model_path, decoder_only=self.fast_encoder)
+        whisper_model = load_model(
+            name=self.model_path if self.model_path else self.model_name,
+            download_root=self.model_path,
+            decoder_only=self.fast_encoder,
+            custom_alignment_heads=self.custom_alignment_heads
+            )
         warmup_audio = load_file(self.warmup_file)
         if warmup_audio is not None:
             warmup_audio = torch.from_numpy(warmup_audio).float()
@@ -329,7 +267,7 @@ class SimulStreamingASR():
             else:
                 # For standard encoder, use the original transcribe warmup
                 warmup_audio = load_file(self.warmup_file)
-                whisper_model.transcribe(warmup_audio, language=self.original_language if self.original_language != 'auto' else None)
+                whisper_model.transcribe(warmup_audio, language=self.lan if self.lan != 'auto' else None)
         return whisper_model
     
     def get_new_model_instance(self):
@@ -362,4 +300,4 @@ class SimulStreamingASR():
         """
         Warmup is done directly in load_model
         """
-        pass
+        pass

whisperlivekit/simul_whisper/config.py

@@ -4,26 +4,22 @@ from dataclasses import dataclass, field
 from typing import Literal
 
 @dataclass
-class SimulWhisperConfig:
-    '''Options that are common for all simul policies that could be implemented in SimulWhisper.'''
-    model_path: str
-    language: str = field(default="zh")
-    nonspeech_prob: float = 0.5
-    audio_min_len: float = 1.0
-    decoder_type: Literal["greedy","beam"] = "greedy"
-    beam_size: int = 5
-    task: Literal["transcribe","translate"] = "transcribe"
-    init_prompt: str = field(default=None)
-    static_init_prompt: str = field(default=None)
-    max_context_tokens: int = field(default=None)
-
-@dataclass
-class AlignAttConfig(SimulWhisperConfig):
-    '''Options specific to the AlignAtt policy.'''
+class AlignAttConfig():
     eval_data_path: str = "tmp"
     segment_length: float = field(default=1.0, metadata = {"help": "in second"})
     frame_threshold: int = 4
     rewind_threshold: int = 200
     audio_max_len: float = 20.0
     cif_ckpt_path: str = ""
-    never_fire: bool = False
+    never_fire: bool = False
+    language: str = field(default="zh")
+    nonspeech_prob: float = 0.5
+    audio_min_len: float = 1.0
+    decoder_type: Literal["greedy","beam"] = "greedy"
+    beam_size: int = 5
+    task: Literal["transcribe","translate"] = "transcribe"
+    tokenizer_is_multilingual: bool = False
+    init_prompt: str = field(default=None)
+    static_init_prompt: str = field(default=None)
+    max_context_tokens: int = field(default=None)
+    

whisperlivekit/simul_whisper/license_simulstreaming.py

@@ -1,5 +0,0 @@
-SIMULSTREAMING_LICENSE = f"""
-SimulStreaming backend is dual-licensed:
-• Non-Commercial Use: PolyForm Noncommercial License 1.0.0.
-• Commercial Use: Check SimulStreaming README (github.com/ufal/SimulStreaming) for more details.
-"""

whisperlivekit/simul_whisper/simul_whisper.py

@@ -8,6 +8,7 @@ import torch.nn.functional as F
 
 from .whisper import load_model, DecodingOptions, tokenizer
 from .config import AlignAttConfig
+from whisperlivekit.timed_objects import ASRToken
 from .whisper.audio import log_mel_spectrogram, TOKENS_PER_SECOND, pad_or_trim, N_SAMPLES, N_FRAMES
 from .whisper.timing import median_filter
 from .whisper.decoding import GreedyDecoder, BeamSearchDecoder, SuppressTokens, detect_language
@@ -18,6 +19,7 @@ from time import time
 from .token_buffer import TokenBuffer
 
 import numpy as np
+from ..timed_objects import PUNCTUATION_MARKS
 from .generation_progress import *
 
 DEC_PAD = 50257
@@ -40,12 +42,6 @@ else:
     except ImportError:
         HAS_FASTER_WHISPER = False
 
-# New features added to the original version of Simul-Whisper: 
-# - large-v3 model support
-# - translation support
-# - beam search
-# - prompt -- static vs. non-static
-# - context
 class PaddedAlignAttWhisper:
     def __init__(
             self, 
@@ -55,30 +51,28 @@ class PaddedAlignAttWhisper:
             fw_encoder=None,
         ) -> None:
         self.log_segments = 0
-        model_name = os.path.basename(cfg.model_path).replace(".pt", "")
-        model_path = os.path.dirname(os.path.abspath(cfg.model_path))
-        if loaded_model:
-            self.model = loaded_model
-        else:
-            self.model = load_model(name=model_name, download_root=model_path)
         
-        self.device = 'cuda' if torch.cuda.is_available() else 'cpu'
-
+        self.model = loaded_model
         self.mlx_encoder = mlx_encoder
         self.fw_encoder = fw_encoder
         if fw_encoder:
             self.fw_feature_extractor = FeatureExtractor(feature_size=self.model.dims.n_mels)
-            
+        
+        self.device = 'cuda' if torch.cuda.is_available() else 'cpu'
+        
         logger.info(f"Model dimensions: {self.model.dims}")
-
+        self.speaker = -1
         self.decode_options = DecodingOptions(
             language = cfg.language, 
             without_timestamps = True,
             task=cfg.task
         )
-        self.tokenizer_is_multilingual = not model_name.endswith(".en")
+        self.tokenizer_is_multilingual = cfg.tokenizer_is_multilingual
         self.create_tokenizer(cfg.language if cfg.language != "auto" else None)
+        # self.create_tokenizer('en')
         self.detected_language = cfg.language if cfg.language != "auto" else None
+        self.global_time_offset = 0.0
+        self.reset_tokenizer_to_auto_next_call = False
         
         self.max_text_len = self.model.dims.n_text_ctx
         self.num_decoder_layers = len(self.model.decoder.blocks)
@@ -153,6 +147,7 @@ class PaddedAlignAttWhisper:
         
         self.last_attend_frame = -self.cfg.rewind_threshold
         self.cumulative_time_offset = 0.0
+        self.first_timestamp = None
 
         if self.cfg.max_context_tokens is None:
             self.max_context_tokens = self.max_text_len
@@ -172,7 +167,10 @@ class PaddedAlignAttWhisper:
             self.inference.kv_cache = self.kv_cache
 
             self.token_decoder = BeamSearchDecoder(inference=self.inference, eot=self.tokenizer.eot, beam_size=cfg.beam_size)
-            
+
+        # Tokens to carry over to next chunk for incomplete UTF-8 characters
+        self.pending_incomplete_tokens = []
+
     def remove_hooks(self):
         for hook in self.l_hooks:
             hook.remove()
@@ -260,13 +258,13 @@ class PaddedAlignAttWhisper:
         self.init_context()
         logger.debug(f"Context: {self.context}")
         if not complete and len(self.segments) > 2:
-            logger.debug("keeping last two segments because they are and it is not complete.")
             self.segments = self.segments[-2:]
         else:
             logger.debug("removing all segments.")
             self.segments = []
         self.log_segments += 1
 
+        self.pending_incomplete_tokens = []
 
     def fire_at_boundary(self, chunked_encoder_feature: torch.Tensor):
         if self.always_fire: return True
@@ -328,7 +326,7 @@ class PaddedAlignAttWhisper:
             self.segments = self.segments[1:]
             logger.debug(f"remove segments: {len(self.segments)} {len(self.tokens)}, cumulative offset: {self.cumulative_time_offset:.2f}s")
             if len(self.tokens) > 1:
-                self.context.append_token_ids(self.tokens[1][0,:])
+                self.context.append_token_ids(self.tokens[1][0,:].tolist())
                 self.tokens = [self.initial_tokens] + self.tokens[2:]
         return removed_len
 
@@ -382,11 +380,11 @@ class PaddedAlignAttWhisper:
         new_segment = True
         if len(self.segments) == 0:
             logger.debug("No segments, nothing to do")
-            return [], {}
+            return []
         if not self._apply_minseglen():
             logger.debug(f"applied minseglen {self.cfg.audio_min_len} > {self.segments_len()}.")
             input_segments = torch.cat(self.segments, dim=0)
-            return [], {}
+            return []
 
         # input_segments is concatenation of audio, it's one array
         if len(self.segments) > 1:
@@ -394,6 +392,13 @@ class PaddedAlignAttWhisper:
         else:
             input_segments = self.segments[0]
 
+        # if self.cfg.language == "auto" and self.reset_tokenizer_to_auto_next_call:
+        #     logger.debug("Resetting tokenizer to auto for new sentence.")
+        #     self.create_tokenizer(None)
+        #     self.detected_language = None
+        #     self.init_tokens()
+        #     self.reset_tokenizer_to_auto_next_call = False
+
         # NEW : we can use a different encoder, before using standart whisper for cross attention with the hooks on the decoder
         beg_encode = time()
         if self.mlx_encoder:
@@ -426,58 +431,38 @@ class PaddedAlignAttWhisper:
         end_encode = time()
         # print('Encoder duration:', end_encode-beg_encode)
                 
-#        logger.debug(f"Encoder feature shape: {encoder_feature.shape}")
-#        if mel.shape[-2:] != (self.model.dims.n_audio_ctx, self.model.dims.n_audio_state):
-#            logger.debug("mel ")
-        if self.cfg.language == "auto" and self.detected_language is None:
-            language_tokens, language_probs = self.lang_id(encoder_feature) 
-            logger.debug(f"Language tokens: {language_tokens}, probs: {language_probs}")
-            top_lan, p = max(language_probs[0].items(), key=lambda x: x[1])
-            logger.info(f"Detected language: {top_lan} with p={p:.4f}")
-            #self.tokenizer.language = top_lan
-            #self.tokenizer.__post_init__()
-            self.create_tokenizer(top_lan)
-            self.detected_language = top_lan
-            self.init_tokens()
-            logger.info(f"Tokenizer language: {self.tokenizer.language}, {self.tokenizer.sot_sequence_including_notimestamps}")
+        if self.cfg.language == "auto" and self.detected_language is None and self.first_timestamp:
+            seconds_since_start = self.segments_len() - self.first_timestamp
+            if seconds_since_start >= 2.0:
+                language_tokens, language_probs = self.lang_id(encoder_feature) 
+                top_lan, p = max(language_probs[0].items(), key=lambda x: x[1])
+                print(f"Detected language: {top_lan} with p={p:.4f}")
+                self.create_tokenizer(top_lan)
+                self.last_attend_frame = -self.cfg.rewind_threshold
+                self.cumulative_time_offset = 0.0
+                self.init_tokens()
+                self.init_context()
+                self.detected_language = top_lan
+                logger.info(f"Tokenizer language: {self.tokenizer.language}, {self.tokenizer.sot_sequence_including_notimestamps}")
 
         self.trim_context()
         current_tokens = self._current_tokens()
-#        
+   
         fire_detected = self.fire_at_boundary(encoder_feature[:, :content_mel_len, :])
 
 
-        ####################### Decoding loop
-        logger.info("Decoding loop starts\n")
-
         sum_logprobs = torch.zeros(self.cfg.beam_size, device=self.device)
         completed = False
+        # punctuation_stop = False
 
         attn_of_alignment_heads = None
         most_attended_frame = None
 
         token_len_before_decoding = current_tokens.shape[1]
         
-        generation_progress = []
-        generation = {
-            "starting_tokens": BeamTokens(current_tokens[0,:].clone(), self.cfg.beam_size),
-            "token_len_before_decoding": token_len_before_decoding,
-            #"fire_detected": fire_detected,
-            "frames_len": content_mel_len,
-            "frames_threshold": 4 if is_last else self.cfg.frame_threshold,
-
-            # to be filled later
-            "logits_starting": None,
-
-            # to be filled later
-            "no_speech_prob": None,
-            "no_speech": False,
-
-            # to be filled in the loop
-            "progress": generation_progress,
-        }
+        l_absolute_timestamps = []
+        
         while not completed and current_tokens.shape[1] < self.max_text_len: # bos is 3 tokens
-            generation_progress_loop = []
 
             if new_segment:
                 tokens_for_logits = current_tokens
@@ -486,50 +471,26 @@ class PaddedAlignAttWhisper:
                 tokens_for_logits = current_tokens[:,-1:]
 
             logits = self.logits(tokens_for_logits, encoder_feature) # B, len(tokens), token dict size
-            if new_segment:
-                generation["logits_starting"] = Logits(logits[:,:,:])
 
             if new_segment and self.tokenizer.no_speech is not None:
                 probs_at_sot = logits[:, self.sot_index, :].float().softmax(dim=-1)
                 no_speech_probs = probs_at_sot[:, self.tokenizer.no_speech].tolist()
-                generation["no_speech_prob"] = no_speech_probs[0]
                 if no_speech_probs[0] > self.cfg.nonspeech_prob:
-                    generation["no_speech"] = True
                     logger.info("no speech, stop")
                     break
 
             logits = logits[:, -1, :] # logits for the last token
-            generation_progress_loop.append(("logits_before_suppress",Logits(logits)))
 
             # supress blank tokens only at the beginning of the segment
             if new_segment:
                 logits[:, self.tokenizer.encode(" ") + [self.tokenizer.eot]] = -np.inf
             new_segment = False
             self.suppress_tokens(logits)
-            #generation_progress_loop.append(("logits_after_suppres",BeamLogits(logits[0,:].clone(), self.cfg.beam_size)))
-            generation_progress_loop.append(("logits_after_suppress",Logits(logits)))
-
             current_tokens, completed = self.token_decoder.update(current_tokens, logits, sum_logprobs)
-            generation_progress_loop.append(("beam_tokens",Tokens(current_tokens[:,-1].clone())))
-            generation_progress_loop.append(("sum_logprobs",sum_logprobs.tolist()))
-            generation_progress_loop.append(("completed",completed))
 
             logger.debug(f"Decoding completed: {completed}, sum_logprobs: {sum_logprobs.tolist()}, tokens: ")
             self.debug_print_tokens(current_tokens)
 
-
-            # if self.decoder_type == "beam":
-            #     logger.debug(f"Finished sequences: {self.token_decoder.finished_sequences}")
-
-            #     logprobs = F.log_softmax(logits.float(), dim=-1)
-            #     idx = 0
-            #     logger.debug(f"Beam search topk: {logprobs[idx].topk(self.cfg.beam_size + 1)}")
-            #     logger.debug(f"Greedy search argmax: {logits.argmax(dim=-1)}")
-            # if completed:
-            #     self.debug_print_tokens(current_tokens)
-
-            #     logger.debug("decode stopped because decoder completed")
-
             attn_of_alignment_heads = [[] for _ in range(self.num_align_heads)]
             for i, attn_mat in enumerate(self.dec_attns):
                 layer_rank = int(i % len(self.model.decoder.blocks))
@@ -548,30 +509,24 @@ class PaddedAlignAttWhisper:
                 t = torch.cat(mat, dim=1)
                 tmp.append(t) 
             attn_of_alignment_heads = torch.stack(tmp, dim=1)
-#            logger.debug(str(attn_of_alignment_heads.shape) + " tttady")
             std, mean = torch.std_mean(attn_of_alignment_heads, dim=-2, keepdim=True, unbiased=False)
             attn_of_alignment_heads = (attn_of_alignment_heads - mean) / std
             attn_of_alignment_heads = median_filter(attn_of_alignment_heads, 7) # from whisper.timing
             attn_of_alignment_heads = attn_of_alignment_heads.mean(dim=1)
-#            logger.debug(str(attn_of_alignment_heads.shape) + " po mean")
             attn_of_alignment_heads = attn_of_alignment_heads[:,:, :content_mel_len]
-#            logger.debug(str(attn_of_alignment_heads.shape) + " pak ")
 
             # for each beam, the most attended frame is:
             most_attended_frames = torch.argmax(attn_of_alignment_heads[:,-1,:], dim=-1)
-            generation_progress_loop.append(("most_attended_frames",most_attended_frames.clone().tolist()))
             
             # Calculate absolute timestamps accounting for cumulative offset
             absolute_timestamps = [(frame * 0.02 + self.cumulative_time_offset) for frame in most_attended_frames.tolist()]
-            generation_progress_loop.append(("absolute_timestamps", absolute_timestamps))
             
             logger.debug(str(most_attended_frames.tolist()) + " most att frames")
             logger.debug(f"Absolute timestamps: {absolute_timestamps} (offset: {self.cumulative_time_offset:.2f}s)")
 
             most_attended_frame = most_attended_frames[0].item()
+            l_absolute_timestamps.append(absolute_timestamps[0])
 
-
-            generation_progress.append(dict(generation_progress_loop))
             logger.debug("current tokens" + str(current_tokens.shape))
             if completed:
             #    # stripping the last token, the eot
@@ -609,66 +564,72 @@ class PaddedAlignAttWhisper:
                     self.tokenizer.decode([current_tokens[i, -1].item()])
                 ))
 
-#        for k,v in generation.items():
-#            print(k,v,file=sys.stderr)
-#        for x in generation_progress:
-#            for y in x.items():
-#                print("\t\t",*y,file=sys.stderr)
-#            print("\t","----", file=sys.stderr)
-#        print("\t", "end of generation_progress_loop", file=sys.stderr)
-        #    sys.exit(1)
-        ####################### End of decoding loop
-
-        logger.info("End of decoding loop")
-
-        # if attn_of_alignment_heads is not None:
-        #     seg_len = int(segment.shape[0] / 16000 * TOKENS_PER_SECOND)
-
-        #     # Lets' now consider only the top hypothesis in the beam search
-        #     top_beam_attn_of_alignment_heads = attn_of_alignment_heads[0]
-
-        #     # debug print: how is the new token attended?
-        #     new_token_attn = top_beam_attn_of_alignment_heads[token_len_before_decoding:, -seg_len:]
-        #     logger.debug(f"New token attention shape: {new_token_attn.shape}")
-        #     if new_token_attn.shape[0] == 0:  # it's not attended in the current audio segment
-        #         logger.debug("no token generated")
-        #     else:  # it is, and the max attention is:
-        #         new_token_max_attn, _ = new_token_attn.max(dim=-1)
-        #         logger.debug(f"segment max attention: {new_token_max_attn.mean().item()/len(self.segments)}")
-
-
-        # let's now operate only with the top beam hypothesis
         tokens_to_split = current_tokens[0, token_len_before_decoding:]
-        if fire_detected or is_last:
+
+        # Prepend pending tokens from previous chunk if any
+        if self.pending_incomplete_tokens:
+            logger.debug(f"[UTF-8 Fix] Prepending {len(self.pending_incomplete_tokens)} pending tokens: {self.pending_incomplete_tokens}")
+            pending_tensor = torch.tensor(self.pending_incomplete_tokens, dtype=torch.long, device=self.device)
+            tokens_to_split = torch.cat([pending_tensor, tokens_to_split])
+
+        if fire_detected or is_last: #or punctuation_stop:
             new_hypothesis = tokens_to_split.flatten().tolist()
+            split_words, split_tokens = self.tokenizer.split_to_word_tokens(new_hypothesis)
         else:
             # going to truncate the tokens after the last space
             split_words, split_tokens = self.tokenizer.split_to_word_tokens(tokens_to_split.tolist())
-            generation["result"] = {"split_words": split_words[:-1], "split_tokens": split_tokens[:-1]}
-            generation["result_truncated"] = {"split_words": split_words[-1:], "split_tokens": split_tokens[-1:]}
-
-#            text_to_split = self.tokenizer.decode(tokens_to_split)
-#            logger.debug(f"text_to_split: {text_to_split}")
-#            logger.debug("text at current step: {}".format(text_to_split.replace(" ", "<space>")))
-#            text_before_space = " ".join(text_to_split.split(" ")[:-1])
-#            logger.debug("before the last space: {}".format(text_before_space.replace(" ", "<space>")))
             if len(split_words) > 1:
                 new_hypothesis = [i for sublist in split_tokens[:-1] for i in sublist]  
             else:
                 new_hypothesis = []
 
 
-        ### new hypothesis
         logger.debug(f"new_hypothesis: {new_hypothesis}")
         new_tokens = torch.tensor([new_hypothesis], dtype=torch.long).repeat_interleave(self.cfg.beam_size, dim=0).to(
             device=self.device,
         )
         self.tokens.append(new_tokens)
-        # TODO: test if this is redundant or not
-#        ret = ret[ret<DEC_PAD]
 
         logger.info(f"Output: {self.tokenizer.decode(new_hypothesis)}")
         
         self._clean_cache()
 
-        return new_hypothesis, generation
+        if len(l_absolute_timestamps) >=2 and self.first_timestamp is None:
+            self.first_timestamp = l_absolute_timestamps[0]
+            
+
+        timestamped_words = []
+        timestamp_idx = 0
+        replacement_char = "\ufffd"
+        for word, word_tokens in zip(split_words, split_tokens):
+            # Skip words containing incomplete UTF-8 from client output
+            if replacement_char in word:
+                logger.warning(f"[UTF-8 Filter] Skipping incomplete word from client output: {repr(word)}")
+                timestamp_idx += len(word_tokens)
+                continue
+
+            try:
+                current_timestamp = l_absolute_timestamps[timestamp_idx]
+            except:
+                pass
+            timestamp_idx += len(word_tokens)
+
+            timestamp_entry = ASRToken(
+                    start=current_timestamp,
+                    end=current_timestamp + 0.1,
+                    text= word,
+                    probability=0.95,
+                    speaker=self.speaker,
+                    detected_language=self.detected_language
+                ).with_offset(
+                    self.global_time_offset
+            )
+            timestamped_words.append(timestamp_entry)
+
+        # Hold incomplete tokens for next chunk
+        self.pending_incomplete_tokens = []
+        if split_words and replacement_char in split_words[-1]:
+            self.pending_incomplete_tokens = split_tokens[-1]
+            logger.warning(f"[UTF-8 Fix] Holding {len(self.pending_incomplete_tokens)} incomplete tokens for next chunk: {self.pending_incomplete_tokens}")
+
+        return timestamped_words

whisperlivekit/simul_whisper/token_buffer.py

@@ -7,6 +7,7 @@ class TokenBuffer:
         self.prefix_token_ids = prefix_token_ids
         self.tokenizer = tokenizer
         self.device = device
+        self.pending_token_ids = []
 
     def as_token_ids(self, tokenizer=None):
 
@@ -64,7 +65,26 @@ class TokenBuffer:
     def append_token_ids(self, token_ids):
         tokenizer = self.tokenizer
         assert tokenizer is not None, "Tokenizer is not set."
-        self.text += self.tokenizer.decode(token_ids)
+
+        all_tokens = self.pending_token_ids + token_ids
+
+        decoded = tokenizer.decode(all_tokens)
+        replacement_char = "\ufffd"
+
+        if replacement_char in decoded:
+            if len(all_tokens) > 1:
+                decoded_partial = tokenizer.decode(all_tokens[:-1])
+
+                if replacement_char not in decoded_partial:
+                    self.text += decoded_partial
+                    self.pending_token_ids = [all_tokens[-1]]
+                else:
+                    self.pending_token_ids = all_tokens
+            else:
+                self.pending_token_ids = all_tokens
+        else:
+            self.text += decoded
+            self.pending_token_ids = []
 
     def as_split_word_tokens(self):
         tokenizer = self.tokenizer

whisperlivekit/simul_whisper/whisper/__init__.py

@@ -105,7 +105,8 @@ def load_model(
     device: Optional[Union[str, torch.device]] = None,
     download_root: str = None,
     in_memory: bool = False,
-    decoder_only=False
+    decoder_only=False,
+    custom_alignment_heads=None
 ) -> Whisper:
     """
     Load a Whisper ASR model
@@ -135,15 +136,17 @@ def load_model(
         download_root = os.path.join(os.getenv("XDG_CACHE_HOME", default), "whisper")
 
     if name in _MODELS:
-        checkpoint_file = _download(_MODELS[name], download_root, in_memory)
-        alignment_heads = _ALIGNMENT_HEADS[name]
+        checkpoint_file = _download(_MODELS[name], download_root, in_memory)        
     elif os.path.isfile(name):
         checkpoint_file = open(name, "rb").read() if in_memory else name
-        alignment_heads = None
     else:
         raise RuntimeError(
             f"Model {name} not found; available models = {available_models()}"
         )
+        
+    alignment_heads = _ALIGNMENT_HEADS.get(name, None)
+    if custom_alignment_heads:
+        alignment_heads = custom_alignment_heads.encode()
 
     with (
         io.BytesIO(checkpoint_file) if in_memory else open(checkpoint_file, "rb")

whisperlivekit/timed_objects.py

@@ -1,7 +1,9 @@
-from dataclasses import dataclass
-from typing import Optional
+from dataclasses import dataclass, field
+from typing import Optional, Any, List
 from datetime import timedelta
 
+PUNCTUATION_MARKS = {'.', '!', '?', '。', '！', '？'}
+
 def format_time(seconds: float) -> str:
     """Format seconds as HH:MM:SS."""
     return str(timedelta(seconds=int(seconds)))
@@ -15,12 +17,41 @@ class TimedText:
     speaker: Optional[int] = -1
     probability: Optional[float] = None
     is_dummy: Optional[bool] = False
+    detected_language: Optional[str] = None
+    
+    def is_punctuation(self):
+        return self.text.strip() in PUNCTUATION_MARKS
+    
+    def overlaps_with(self, other: 'TimedText') -> bool:
+        return not (self.end <= other.start or other.end <= self.start)
+    
+    def is_within(self, other: 'TimedText') -> bool:
+        return other.contains_timespan(self)
 
-@dataclass
+    def duration(self) -> float:
+        return self.end - self.start
+
+    def contains_time(self, time: float) -> bool:
+        return self.start <= time <= self.end
+
+    def contains_timespan(self, other: 'TimedText') -> bool:
+        return self.start <= other.start and self.end >= other.end
+    
+    def __bool__(self):
+        return bool(self.text)
+
+
+@dataclass()
 class ASRToken(TimedText):
+    
+    corrected_speaker: Optional[int] = -1
+    validated_speaker: bool = False
+    validated_text: bool = False
+    validated_language: bool = False
+    
     def with_offset(self, offset: float) -> "ASRToken":
         """Return a new token with the time offset added."""
-        return ASRToken(self.start + offset, self.end + offset, self.text, self.speaker, self.probability)
+        return ASRToken(self.start + offset, self.end + offset, self.text, self.speaker, self.probability, detected_language=self.detected_language)
 
 @dataclass
 class Sentence(TimedText):
@@ -28,7 +59,28 @@ class Sentence(TimedText):
 
 @dataclass
 class Transcript(TimedText):
-    pass
+    """
+    represents a concatenation of several ASRToken
+    """
+
+    @classmethod
+    def from_tokens(
+        cls,
+        tokens: List[ASRToken],
+        sep: Optional[str] = None,
+        offset: float = 0
+    ) -> "Transcript":
+        sep = sep if sep is not None else ' '
+        text = sep.join(token.text for token in tokens)
+        probability = sum(token.probability for token in tokens if token.probability) / len(tokens) if tokens else None
+        if tokens:
+            start = offset + tokens[0].start
+            end = offset + tokens[-1].end
+        else:
+            start = None
+            end = None
+        return cls(start, end, text, probability=probability)
+
 
 @dataclass
 class SpeakerSegment(TimedText):
@@ -39,8 +91,53 @@ class SpeakerSegment(TimedText):
 
 @dataclass
 class Translation(TimedText):
+    is_validated : bool = False
     pass
 
+    # def split(self):
+    #     return self.text.split(" ") # should be customized with the sep
+
+    def approximate_cut_at(self, cut_time):
+        """
+        Each word in text is considered to be of duration (end-start)/len(words in text)
+        """
+        if not self.text or not self.contains_time(cut_time):
+            return self, None
+
+        words = self.text.split()
+        num_words = len(words)
+        if num_words == 0:
+            return self, None
+
+        duration_per_word = self.duration() / num_words
+        
+        cut_word_index = int((cut_time - self.start) / duration_per_word)
+        
+        if cut_word_index >= num_words:
+            cut_word_index = num_words -1
+        
+        text0 = " ".join(words[:cut_word_index])
+        text1 = " ".join(words[cut_word_index:])
+
+        segment0 = Translation(start=self.start, end=cut_time, text=text0)
+        segment1 = Translation(start=cut_time, end=self.end, text=text1)
+
+        return segment0, segment1
+        
+    def cut_position(self, position):
+        sep=" "
+        words = self.text.split(sep)
+        num_words = len(words)
+        duration_per_word = self.duration() / num_words
+        cut_time=duration_per_word*position
+        
+        text0 = sep.join(words[:position])
+        text1 = sep.join(words[position:])
+
+        segment0 = Translation(start=self.start, end=cut_time, text=text0)
+        segment1 = Translation(start=cut_time, end=self.end, text=text1)
+        return segment0, segment1
+
 @dataclass
 class Silence():
     duration: float
@@ -51,10 +148,142 @@ class Line(TimedText):
     translation: str = ''
     
     def to_dict(self):
-        return {
-            'speaker': int(self.speaker),
+        _dict = {
+            'speaker': int(self.speaker) if self.speaker != -1 else 1,
             'text': self.text,
-            'translation': self.translation,
             'start': format_time(self.start),
             'end': format_time(self.end),
-        }
+        }
+        if self.translation:
+            _dict['translation'] = self.translation
+        if self.detected_language:
+            _dict['detected_language'] = self.detected_language
+        return _dict
+    
+@dataclass
+class WordValidation:
+    """Validation status for word-level data."""
+    text: bool = False
+    speaker: bool = False
+    language: bool = False
+    
+    def to_dict(self):
+        return {
+            'text': self.text,
+            'speaker': self.speaker,
+            'language': self.language
+        }
+
+
+@dataclass
+class Word:
+    """Word-level object with timing and validation information."""
+    text: str = ''
+    start: float = 0.0
+    end: float = 0.0
+    validated: WordValidation = field(default_factory=WordValidation)
+    
+    def to_dict(self):
+        return {
+            'text': self.text,
+            'start': self.start,
+            'end': self.end,
+            'validated': self.validated.to_dict()
+        }
+
+
+@dataclass
+class SegmentBuffer:
+    """Per-segment temporary buffers for ephemeral data."""
+    transcription: str = ''
+    diarization: str = ''
+    translation: str = ''
+    
+    def to_dict(self):
+        return {
+            'transcription': self.transcription,
+            'diarization': self.diarization,
+            'translation': self.translation
+        }
+
+
+@dataclass
+class Segment:
+    """Represents a segment in the new API structure."""
+    id: int = 0
+    speaker: int = -1
+    text: str = ''
+    start_speaker: float = 0.0
+    start: float = 0.0
+    end: float = 0.0
+    language: Optional[str] = None
+    translation: str = ''
+    words: List[ASRToken] = field(default_factory=list)
+    buffer_tokens: List[ASRToken] = field(default_factory=list)
+    buffer_translation = ''
+    buffer: SegmentBuffer = field(default_factory=SegmentBuffer)
+    
+    def to_dict(self):
+        """Convert segment to dictionary for JSON serialization."""
+        return {
+            'id': self.id,
+            'speaker': self.speaker,
+            'text': self.text,
+            'start_speaker': self.start_speaker,
+            'start': self.start,
+            'end': self.end,
+            'language': self.language,
+            'translation': self.translation,
+            'words': [word.to_dict() for word in self.words],
+            'buffer': self.buffer.to_dict()
+        }
+
+    def consolidate(self, sep):
+        self.text = sep.join([word.text for word in self.words])
+        if self.words:
+            self.start = self.words[0].start
+            self.end = self.words[-1].end
+        
+
+@dataclass  
+class FrontData():
+    status: str = ''
+    error: str = ''
+    lines: list[Line] = field(default_factory=list)
+    buffer_transcription: str = ''
+    buffer_diarization: str = ''
+    remaining_time_transcription: float = 0.
+    remaining_time_diarization: float = 0.
+    
+    def to_dict(self):
+        _dict = {
+            'status': self.status,
+            'lines': [line.to_dict() for line in self.lines if (line.text or line.speaker == -2)],
+            'buffer_transcription': self.buffer_transcription,
+            'buffer_diarization': self.buffer_diarization,
+            'remaining_time_transcription': self.remaining_time_transcription,
+            'remaining_time_diarization': self.remaining_time_diarization,
+        }
+        if self.error:
+            _dict['error'] = self.error
+        return _dict
+
+@dataclass  
+class ChangeSpeaker:
+    speaker: int
+    start: int
+
+@dataclass  
+class State():
+    tokens: list = field(default_factory=list)
+    segments: list = field(default_factory=list)
+    last_validated_token: int = 0
+    last_validated_segment: int = 0 # validated means tokens speaker and transcription are validated and terminated
+    translation_validated_segments: list = field(default_factory=list)
+    translation_buffer: list = field(default_factory=list)
+    buffer_transcription: str = field(default_factory=Transcript)
+    end_buffer: float = 0.0
+    end_attributed_speaker: float = 0.0
+    remaining_time_transcription: float = 0.0
+    remaining_time_diarization: float = 0.0
+    beg_loop: Optional[int] = None

whisperlivekit/translation/mapping_languages.py

@@ -1,182 +0,0 @@
-"""
-adapted from https://store.crowdin.com/custom-mt
-"""
-
-LANGUAGES = [
-    {"name": "Afrikaans", "nllb": "afr_Latn", "crowdin": "af"},
-    {"name": "Akan", "nllb": "aka_Latn", "crowdin": "ak"},
-    {"name": "Amharic", "nllb": "amh_Ethi", "crowdin": "am"},
-    {"name": "Assamese", "nllb": "asm_Beng", "crowdin": "as"},
-    {"name": "Asturian", "nllb": "ast_Latn", "crowdin": "ast"},
-    {"name": "Bashkir", "nllb": "bak_Cyrl", "crowdin": "ba"},
-    {"name": "Bambara", "nllb": "bam_Latn", "crowdin": "bm"},
-    {"name": "Balinese", "nllb": "ban_Latn", "crowdin": "ban"},
-    {"name": "Belarusian", "nllb": "bel_Cyrl", "crowdin": "be"},
-    {"name": "Bengali", "nllb": "ben_Beng", "crowdin": "bn"},
-    {"name": "Bosnian", "nllb": "bos_Latn", "crowdin": "bs"},
-    {"name": "Bulgarian", "nllb": "bul_Cyrl", "crowdin": "bg"},
-    {"name": "Catalan", "nllb": "cat_Latn", "crowdin": "ca"},
-    {"name": "Cebuano", "nllb": "ceb_Latn", "crowdin": "ceb"},
-    {"name": "Czech", "nllb": "ces_Latn", "crowdin": "cs"},
-    {"name": "Welsh", "nllb": "cym_Latn", "crowdin": "cy"},
-    {"name": "Danish", "nllb": "dan_Latn", "crowdin": "da"},
-    {"name": "German", "nllb": "deu_Latn", "crowdin": "de"},
-    {"name": "Dzongkha", "nllb": "dzo_Tibt", "crowdin": "dz"},
-    {"name": "Greek", "nllb": "ell_Grek", "crowdin": "el"},
-    {"name": "English", "nllb": "eng_Latn", "crowdin": "en"},
-    {"name": "Esperanto", "nllb": "epo_Latn", "crowdin": "eo"},
-    {"name": "Estonian", "nllb": "est_Latn", "crowdin": "et"},
-    {"name": "Basque", "nllb": "eus_Latn", "crowdin": "eu"},
-    {"name": "Ewe", "nllb": "ewe_Latn", "crowdin": "ee"},
-    {"name": "Faroese", "nllb": "fao_Latn", "crowdin": "fo"},
-    {"name": "Fijian", "nllb": "fij_Latn", "crowdin": "fj"},
-    {"name": "Finnish", "nllb": "fin_Latn", "crowdin": "fi"},
-    {"name": "French", "nllb": "fra_Latn", "crowdin": "fr"},
-    {"name": "Friulian", "nllb": "fur_Latn", "crowdin": "fur-IT"},
-    {"name": "Scottish Gaelic", "nllb": "gla_Latn", "crowdin": "gd"},
-    {"name": "Irish", "nllb": "gle_Latn", "crowdin": "ga-IE"},
-    {"name": "Galician", "nllb": "glg_Latn", "crowdin": "gl"},
-    {"name": "Guarani", "nllb": "grn_Latn", "crowdin": "gn"},
-    {"name": "Gujarati", "nllb": "guj_Gujr", "crowdin": "gu-IN"},
-    {"name": "Haitian Creole", "nllb": "hat_Latn", "crowdin": "ht"},
-    {"name": "Hausa", "nllb": "hau_Latn", "crowdin": "ha"},
-    {"name": "Hebrew", "nllb": "heb_Hebr", "crowdin": "he"},
-    {"name": "Hindi", "nllb": "hin_Deva", "crowdin": "hi"},
-    {"name": "Croatian", "nllb": "hrv_Latn", "crowdin": "hr"},
-    {"name": "Hungarian", "nllb": "hun_Latn", "crowdin": "hu"},
-    {"name": "Armenian", "nllb": "hye_Armn", "crowdin": "hy-AM"},
-    {"name": "Igbo", "nllb": "ibo_Latn", "crowdin": "ig"},
-    {"name": "Indonesian", "nllb": "ind_Latn", "crowdin": "id"},
-    {"name": "Icelandic", "nllb": "isl_Latn", "crowdin": "is"},
-    {"name": "Italian", "nllb": "ita_Latn", "crowdin": "it"},
-    {"name": "Javanese", "nllb": "jav_Latn", "crowdin": "jv"},
-    {"name": "Japanese", "nllb": "jpn_Jpan", "crowdin": "ja"},
-    {"name": "Kabyle", "nllb": "kab_Latn", "crowdin": "kab"},
-    {"name": "Kannada", "nllb": "kan_Knda", "crowdin": "kn"},
-    {"name": "Georgian", "nllb": "kat_Geor", "crowdin": "ka"},
-    {"name": "Kazakh", "nllb": "kaz_Cyrl", "crowdin": "kk"},
-    {"name": "Khmer", "nllb": "khm_Khmr", "crowdin": "km"},
-    {"name": "Kinyarwanda", "nllb": "kin_Latn", "crowdin": "rw"},
-    {"name": "Kyrgyz", "nllb": "kir_Cyrl", "crowdin": "ky"},
-    {"name": "Korean", "nllb": "kor_Hang", "crowdin": "ko"},
-    {"name": "Lao", "nllb": "lao_Laoo", "crowdin": "lo"},
-    {"name": "Ligurian", "nllb": "lij_Latn", "crowdin": "lij"},
-    {"name": "Limburgish", "nllb": "lim_Latn", "crowdin": "li"},
-    {"name": "Lingala", "nllb": "lin_Latn", "crowdin": "ln"},
-    {"name": "Lithuanian", "nllb": "lit_Latn", "crowdin": "lt"},
-    {"name": "Luxembourgish", "nllb": "ltz_Latn", "crowdin": "lb"},
-    {"name": "Maithili", "nllb": "mai_Deva", "crowdin": "mai"},
-    {"name": "Malayalam", "nllb": "mal_Mlym", "crowdin": "ml-IN"},
-    {"name": "Marathi", "nllb": "mar_Deva", "crowdin": "mr"},
-    {"name": "Macedonian", "nllb": "mkd_Cyrl", "crowdin": "mk"},
-    {"name": "Maltese", "nllb": "mlt_Latn", "crowdin": "mt"},
-    {"name": "Mossi", "nllb": "mos_Latn", "crowdin": "mos"},
-    {"name": "Maori", "nllb": "mri_Latn", "crowdin": "mi"},
-    {"name": "Burmese", "nllb": "mya_Mymr", "crowdin": "my"},
-    {"name": "Dutch", "nllb": "nld_Latn", "crowdin": "nl"},
-    {"name": "Norwegian Nynorsk", "nllb": "nno_Latn", "crowdin": "nn-NO"},
-    {"name": "Nepali", "nllb": "npi_Deva", "crowdin": "ne-NP"},
-    {"name": "Northern Sotho", "nllb": "nso_Latn", "crowdin": "nso"},
-    {"name": "Occitan", "nllb": "oci_Latn", "crowdin": "oc"},
-    {"name": "Odia", "nllb": "ory_Orya", "crowdin": "or"},
-    {"name": "Papiamento", "nllb": "pap_Latn", "crowdin": "pap"},
-    {"name": "Polish", "nllb": "pol_Latn", "crowdin": "pl"},
-    {"name": "Portuguese", "nllb": "por_Latn", "crowdin": "pt-PT"},
-    {"name": "Dari", "nllb": "prs_Arab", "crowdin": "fa-AF"},
-    {"name": "Romanian", "nllb": "ron_Latn", "crowdin": "ro"},
-    {"name": "Rundi", "nllb": "run_Latn", "crowdin": "rn"},
-    {"name": "Russian", "nllb": "rus_Cyrl", "crowdin": "ru"},
-    {"name": "Sango", "nllb": "sag_Latn", "crowdin": "sg"},
-    {"name": "Sanskrit", "nllb": "san_Deva", "crowdin": "sa"},
-    {"name": "Santali", "nllb": "sat_Olck", "crowdin": "sat"},
-    {"name": "Sinhala", "nllb": "sin_Sinh", "crowdin": "si-LK"},
-    {"name": "Slovak", "nllb": "slk_Latn", "crowdin": "sk"},
-    {"name": "Slovenian", "nllb": "slv_Latn", "crowdin": "sl"},
-    {"name": "Shona", "nllb": "sna_Latn", "crowdin": "sn"},
-    {"name": "Sindhi", "nllb": "snd_Arab", "crowdin": "sd"},
-    {"name": "Somali", "nllb": "som_Latn", "crowdin": "so"},
-    {"name": "Southern Sotho", "nllb": "sot_Latn", "crowdin": "st"},
-    {"name": "Spanish", "nllb": "spa_Latn", "crowdin": "es-ES"},
-    {"name": "Sardinian", "nllb": "srd_Latn", "crowdin": "sc"},
-    {"name": "Swati", "nllb": "ssw_Latn", "crowdin": "ss"},
-    {"name": "Sundanese", "nllb": "sun_Latn", "crowdin": "su"},
-    {"name": "Swedish", "nllb": "swe_Latn", "crowdin": "sv-SE"},
-    {"name": "Swahili", "nllb": "swh_Latn", "crowdin": "sw"},
-    {"name": "Tamil", "nllb": "tam_Taml", "crowdin": "ta"},
-    {"name": "Tatar", "nllb": "tat_Cyrl", "crowdin": "tt-RU"},
-    {"name": "Telugu", "nllb": "tel_Telu", "crowdin": "te"},
-    {"name": "Tajik", "nllb": "tgk_Cyrl", "crowdin": "tg"},
-    {"name": "Tagalog", "nllb": "tgl_Latn", "crowdin": "tl"},
-    {"name": "Thai", "nllb": "tha_Thai", "crowdin": "th"},
-    {"name": "Tigrinya", "nllb": "tir_Ethi", "crowdin": "ti"},
-    {"name": "Tswana", "nllb": "tsn_Latn", "crowdin": "tn"},
-    {"name": "Tsonga", "nllb": "tso_Latn", "crowdin": "ts"},
-    {"name": "Turkmen", "nllb": "tuk_Latn", "crowdin": "tk"},
-    {"name": "Turkish", "nllb": "tur_Latn", "crowdin": "tr"},
-    {"name": "Uyghur", "nllb": "uig_Arab", "crowdin": "ug"},
-    {"name": "Ukrainian", "nllb": "ukr_Cyrl", "crowdin": "uk"},
-    {"name": "Venetian", "nllb": "vec_Latn", "crowdin": "vec"},
-    {"name": "Vietnamese", "nllb": "vie_Latn", "crowdin": "vi"},
-    {"name": "Wolof", "nllb": "wol_Latn", "crowdin": "wo"},
-    {"name": "Xhosa", "nllb": "xho_Latn", "crowdin": "xh"},
-    {"name": "Yoruba", "nllb": "yor_Latn", "crowdin": "yo"},
-    {"name": "Zulu", "nllb": "zul_Latn", "crowdin": "zu"},
-]
-
-NAME_TO_NLLB = {lang["name"]: lang["nllb"] for lang in LANGUAGES}
-NAME_TO_CROWDIN = {lang["name"]: lang["crowdin"] for lang in LANGUAGES}
-CROWDIN_TO_NLLB = {lang["crowdin"]: lang["nllb"] for lang in LANGUAGES}
-NLLB_TO_CROWDIN = {lang["nllb"]: lang["crowdin"] for lang in LANGUAGES}
-CROWDIN_TO_NAME = {lang["crowdin"]: lang["name"] for lang in LANGUAGES}
-NLLB_TO_NAME = {lang["nllb"]: lang["name"] for lang in LANGUAGES}
-
-
-def get_nllb_code(crowdin_code):
-    return CROWDIN_TO_NLLB.get(crowdin_code, None)
-
-
-def get_crowdin_code(nllb_code):
-    return NLLB_TO_CROWDIN.get(nllb_code)
-
-
-def get_language_name_by_crowdin(crowdin_code):
-    return CROWDIN_TO_NAME.get(crowdin_code)
-
-
-def get_language_name_by_nllb(nllb_code):
-    return NLLB_TO_NAME.get(nllb_code)
-
-
-def get_language_info(identifier, identifier_type="auto"):
-    if identifier_type == "auto":
-        for lang in LANGUAGES:
-            if (lang["name"].lower() == identifier.lower() or 
-                lang["nllb"] == identifier or 
-                lang["crowdin"] == identifier):
-                return lang
-    elif identifier_type == "name":
-        for lang in LANGUAGES:
-            if lang["name"].lower() == identifier.lower():
-                return lang
-    elif identifier_type == "nllb":
-        for lang in LANGUAGES:
-            if lang["nllb"] == identifier:
-                return lang
-    elif identifier_type == "crowdin":
-        for lang in LANGUAGES:
-            if lang["crowdin"] == identifier:
-                return lang
-    
-    return None
-
-
-def list_all_languages():
-    return [lang["name"] for lang in LANGUAGES]
-
-
-def list_all_nllb_codes():
-    return [lang["nllb"] for lang in LANGUAGES]
-
-
-def list_all_crowdin_codes():
-    return [lang["crowdin"] for lang in LANGUAGES]

whisperlivekit/translation/translation.py

@@ -1,137 +0,0 @@
-import ctranslate2
-import torch
-import transformers
-from dataclasses import dataclass
-import huggingface_hub
-from whisperlivekit.translation.mapping_languages import get_nllb_code
-from whisperlivekit.timed_objects import Translation
-
-
-#In diarization case, we may want to translate just one speaker, or at least start the sentences there
-
-PUNCTUATION_MARKS = {'.', '!', '?', '。', '！', '？'}
-
-
-@dataclass
-class TranslationModel():
-    translator: ctranslate2.Translator
-    tokenizer: dict
-
-def load_model(src_langs):
-    MODEL = 'nllb-200-distilled-600M-ctranslate2'
-    MODEL_GUY = 'entai2965'
-    huggingface_hub.snapshot_download(MODEL_GUY + '/' + MODEL,local_dir=MODEL)
-    device = "cuda" if torch.cuda.is_available() else "cpu"
-    translator = ctranslate2.Translator(MODEL,device=device)
-    tokenizer = dict()
-    for src_lang in src_langs:
-        tokenizer[src_lang] = transformers.AutoTokenizer.from_pretrained(MODEL, src_lang=src_lang, clean_up_tokenization_spaces=True)
-    return TranslationModel(
-        translator=translator,
-        tokenizer=tokenizer
-    )
-
-def translate(input, translation_model, tgt_lang):
-    source = translation_model.tokenizer.convert_ids_to_tokens(translation_model.tokenizer.encode(input))
-    target_prefix = [tgt_lang]
-    results = translation_model.translator.translate_batch([source], target_prefix=[target_prefix])
-    target = results[0].hypotheses[0][1:]
-    return translation_model.tokenizer.decode(translation_model.tokenizer.convert_tokens_to_ids(target))
-
-class OnlineTranslation:
-    def __init__(self, translation_model: TranslationModel, input_languages: list, output_languages: list):
-        self.buffer = []
-        self.len_processed_buffer = 0
-        self.translation_remaining = Translation()
-        self.validated = []
-        self.translation_pending_validation = ''
-        self.translation_model = translation_model
-        self.input_languages = input_languages
-        self.output_languages = output_languages
-
-    def compute_common_prefix(self, results):
-        #we dont want want to prune the result for the moment. 
-        if not self.buffer:
-            self.buffer = results
-        else:
-            for i in range(min(len(self.buffer), len(results))):
-                if self.buffer[i] != results[i]:
-                    self.commited.extend(self.buffer[:i])
-                    self.buffer = results[i:]
-
-    def translate(self, input, input_lang=None, output_lang=None):
-        if not input:
-            return ""
-        if input_lang is None:
-            input_lang = self.input_languages[0]
-        if output_lang is None:
-            output_lang = self.output_languages[0]
-        nllb_output_lang = get_nllb_code(output_lang)
-            
-        source = self.translation_model.tokenizer[input_lang].convert_ids_to_tokens(self.translation_model.tokenizer[input_lang].encode(input))   
-        results = self.translation_model.translator.translate_batch([source], target_prefix=[[nllb_output_lang]]) #we can use return_attention=True to try to optimize the stuff.
-        target = results[0].hypotheses[0][1:]
-        results = self.translation_model.tokenizer[input_lang].decode(self.translation_model.tokenizer[input_lang].convert_tokens_to_ids(target))
-        return results
-
-    def translate_tokens(self, tokens):
-        if tokens:
-            text = ' '.join([token.text for token in tokens])
-            start = tokens[0].start
-            end = tokens[-1].end
-            translated_text = self.translate(text)
-            translation = Translation(
-                text=translated_text,
-                start=start,
-                end=end,
-            )
-            return translation
-        return None
-            
-        
-
-    def insert_tokens(self, tokens):
-        self.buffer.extend(tokens)
-        pass
-    
-    def process(self):
-        i = 0
-        if len(self.buffer) < self.len_processed_buffer + 3: #nothing new to process
-            return self.validated + [self.translation_remaining]
-        while i < len(self.buffer):
-            if self.buffer[i].text in PUNCTUATION_MARKS:
-                translation_sentence = self.translate_tokens(self.buffer[:i+1])
-                self.validated.append(translation_sentence)
-                self.buffer = self.buffer[i+1:]
-                i = 0
-            else:
-                i+=1
-        self.translation_remaining = self.translate_tokens(self.buffer)
-        self.len_processed_buffer = len(self.buffer)
-        return self.validated + [self.translation_remaining]
-                
-
-if __name__ == '__main__':
-    output_lang = 'fr'
-    input_lang = "en"
-    
-    
-    test_string = """
-    Transcription technology has improved so much in the past few years. Have you noticed how accurate real-time speech-to-text is now?
-    """
-    test = test_string.split(' ')
-    step = len(test) // 3
-    
-    shared_model = load_model([input_lang])
-    online_translation = OnlineTranslation(shared_model, input_languages=[input_lang], output_languages=[output_lang])
-        
-    for id in range(5):
-        val = test[id*step : (id+1)*step]
-        val_str = ' '.join(val)
-        result = online_translation.translate(val_str)
-        print(result)
-    
-    
-    
-    
-    # print(result)

whisperlivekit/warmup.py

@@ -6,57 +6,46 @@ logger = logging.getLogger(__name__)
 def load_file(warmup_file=None, timeout=5):
     import os
     import tempfile
+    import urllib.request
     import librosa
-        
+
+    if warmup_file == "":
+        logger.info(f"Skipping warmup.")
+        return None
+
+    # Download JFK sample if not already present
     if warmup_file is None:
-        # Download JFK sample if not already present
         jfk_url = "https://github.com/ggerganov/whisper.cpp/raw/master/samples/jfk.wav"
         temp_dir = tempfile.gettempdir()
         warmup_file = os.path.join(temp_dir, "whisper_warmup_jfk.wav")
-        
-        if not os.path.exists(warmup_file):
-            logger.debug(f"Downloading warmup file from {jfk_url}")
-            print(f"Downloading warmup file from {jfk_url}")
-            import time
-            import urllib.request
-            import urllib.error
-            import socket
-            
-            original_timeout = socket.getdefaulttimeout()
-            socket.setdefaulttimeout(timeout)
-            
-            start_time = time.time()
+        if not os.path.exists(warmup_file) or os.path.getsize(warmup_file) == 0:
             try:
-                urllib.request.urlretrieve(jfk_url, warmup_file)
-                logger.debug(f"Download successful in {time.time() - start_time:.2f}s")
-            except (urllib.error.URLError, socket.timeout) as e:
-                logger.warning(f"Download failed: {e}. Proceeding without warmup.")
+                logger.debug(f"Downloading warmup file from {jfk_url}")
+                with urllib.request.urlopen(jfk_url, timeout=timeout) as r, open(warmup_file, "wb") as f:
+                    f.write(r.read())
+            except Exception as e:
+                logger.warning(f"Warmup file download failed: {e}.")
                 return None
-            finally:
-                socket.setdefaulttimeout(original_timeout)
-    elif not warmup_file:
-        return None 
-    
-    if not warmup_file or not os.path.exists(warmup_file) or os.path.getsize(warmup_file) == 0:
-        logger.warning(f"Warmup file {warmup_file} invalid or missing.")
+
+    # Validate file and load
+    if not os.path.exists(warmup_file) or os.path.getsize(warmup_file) == 0:
+        logger.warning(f"Warmup file {warmup_file} is invalid or missing.")
         return None
-    
+
     try:
-        audio, sr = librosa.load(warmup_file, sr=16000)
+        audio, _ = librosa.load(warmup_file, sr=16000)
+        return audio
     except Exception as e:
-        logger.warning(f"Failed to load audio file: {e}")
+        logger.warning(f"Failed to load warmup file: {e}")
         return None
-    return audio
 
 def warmup_asr(asr, warmup_file=None, timeout=5):
     """
     Warmup the ASR model by transcribing a short audio file.
     """
-    audio = load_file(warmup_file=None, timeout=5)
+    audio = load_file(warmup_file=warmup_file, timeout=timeout)
+    if audio is None:
+        logger.warning("Warmup file unavailable. Skipping ASR warmup.")
+        return
     asr.transcribe(audio)
-    logger.info("ASR model is warmed up")
-    
-def warmup_online(online, warmup_file=None, timeout=5):
-    audio = load_file(warmup_file=None, timeout=5)
-    online.warmup(audio)
-    logger.warning("ASR is warmed up")
+    logger.info("ASR model is warmed up.")

whisperlivekit/web/live_transcription.css

@@ -72,6 +72,12 @@
   --label-trans-text: #111111;
 }
 
+html.is-extension
+{
+    width: 350px;
+    height: 500px;
+}
+
 body {
   font-family: ui-sans-serif, system-ui, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol', 'Noto Color Emoji';
   margin: 0;
@@ -191,6 +197,14 @@ body {
   justify-content: center;
   align-items: center;
   gap: 15px;
+  position: relative;
+  flex-wrap: wrap;
+}
+
+.buttons-container {
+  display: flex;
+  align-items: center;
+  gap: 15px;
 }
 
 .settings {
@@ -200,6 +214,66 @@ body {
   gap: 12px;
 }
 
+.settings-toggle {
+  width: 40px;
+  height: 40px;
+  border: none;
+  border-radius: 50%;
+  background-color: var(--button-bg);
+  border: 1px solid var(--button-border);
+  cursor: pointer;
+  display: none;
+  align-items: center;
+  justify-content: center;
+  transition: all 0.2s ease;
+}
+
+.settings-toggle:hover {
+  background-color: var(--chip-bg);
+}
+
+.settings-toggle.active {
+  background-color: var(--chip-bg);
+}
+
+.settings-toggle img {
+  width: 20px;
+  height: 20px;
+}
+
+@media (max-width: 10000px) {
+  .settings-toggle {
+    display: flex;
+  }
+  
+  .settings {
+    display: none;
+    background: var(--bg);
+    border: 1px solid var(--border);
+    border-radius: 18px;
+    padding: 12px;
+  }
+  
+  .settings.visible {
+    display: flex;
+  }
+}
+
+@media (max-width: 600px) {
+  .settings-container {
+    flex-direction: column;
+    align-items: center;
+    gap: 10px;
+  }
+  
+  .buttons-container {
+    display: flex;
+    justify-content: center;
+    align-items: center;
+    gap: 15px;
+  }
+}
+
 .field {
   display: flex;
   flex-direction: column;
@@ -346,7 +420,7 @@ label {
 
 .label_diarization {
   background-color: var(--chip-bg);
-  border-radius: 8px 8px 8px 8px;
+  border-radius: 100px;
   padding: 2px 10px;
   margin-left: 10px;
   display: inline-block;
@@ -358,7 +432,7 @@ label {
 
 .label_transcription {
   background-color: var(--chip-bg);
-  border-radius: 8px 8px 8px 8px;
+  border-radius: 100px;
   padding: 2px 10px;
   display: inline-block;
   white-space: nowrap;
@@ -370,16 +444,20 @@ label {
 
 .label_translation {
   background-color: var(--chip-bg);
+  display: inline-flex;
   border-radius: 10px;
   padding: 4px 8px;
   margin-top: 4px;
   font-size: 14px;
   color: var(--text);
-  display: flex;
   align-items: flex-start;
   gap: 4px;
 }
 
+.lag-diarization-value {
+    margin-left: 10px;
+}
+
 .label_translation img {
   margin-top: 2px;
 }
@@ -391,7 +469,7 @@ label {
 
 #timeInfo {
   color: var(--muted);
-  margin-left: 10px;
+  margin-left: 0px;
 }
 
 .textcontent {
@@ -405,7 +483,6 @@ label {
 
 .buffer_diarization {
   color: var(--label-dia-text);
-  margin-left: 4px;
 }
 
 .buffer_transcription {
@@ -438,7 +515,6 @@ label {
   font-size: 13px;
   border-radius: 30px;
   padding: 2px 10px;
-  display: none;
 }
 
 .loading {
@@ -451,7 +527,7 @@ label {
 }
 
 /* for smaller screens */
-@media (max-width: 768px) {
+@media (max-width: 200px) {
   .header-container {
     padding: 15px;
   }
@@ -461,6 +537,10 @@ label {
     gap: 10px;
   }
   
+  .buttons-container {
+    gap: 10px;
+  }
+  
   .settings {
     justify-content: center;
     gap: 8px;
@@ -515,3 +595,31 @@ label {
     padding: 10px;
   }
 }
+
+.label_language {
+  background-color: var(--chip-bg);
+  margin-bottom: 0px;
+  border-radius: 100px;
+  padding: 2px 8px;
+  margin-left: 10px;
+  display: inline-flex;
+  align-items: center;
+  gap: 4px;
+  font-size: 14px;
+  color: var(--muted);
+}
+
+
+.speaker-badge {
+  display: inline-flex;
+  align-items: center;
+  justify-content: center;
+  width: 16px;
+  height: 16px;
+  margin-left: -5px;
+  border-radius: 50%;
+  font-size: 11px;
+  line-height: 1;
+  font-weight: 800;
+  color: var(--muted);
+}

whisperlivekit/web/live_transcription.html

@@ -5,23 +5,29 @@
     <meta charset="UTF-8" />
     <meta name="viewport" content="width=device-width, initial-scale=1.0" />
     <title>WhisperLiveKit</title>
-    <link rel="stylesheet" href="/web/live_transcription.css" />
+    <link rel="stylesheet" href="live_transcription.css" />
 </head>
 
 <body>
     <div class="header-container">
         <div class="settings-container">
-            <button id="recordButton">
-                <div class="shape-container">
-                    <div class="shape"></div>
-                </div>
-                <div class="recording-info">
-                    <div class="wave-container">
-                        <canvas id="waveCanvas"></canvas>
+            <div class="buttons-container">
+                <button id="recordButton">
+                    <div class="shape-container">
+                        <div class="shape"></div>
                     </div>
-                    <div class="timer">00:00</div>
-                </div>
-            </button>
+                    <div class="recording-info">
+                        <div class="wave-container">
+                            <canvas id="waveCanvas"></canvas>
+                        </div>
+                        <div class="timer">00:00</div>
+                    </div>
+                </button>
+
+                <button id="settingsToggle" class="settings-toggle" title="Show/hide settings">
+                    <img src="web/src/settings.svg" alt="Settings" />
+                </button>
+            </div>
 
             <div class="settings">
                 <div class="field">
@@ -67,7 +73,7 @@
         <div id="linesTranscript"></div>
     </div>
 
-    <script src="/web/live_transcription.js"></script>
+    <script src="live_transcription.js"></script>
 </body>
 
-</html>
+</html>

whisperlivekit/web/live_transcription.js

@@ -1,4 +1,8 @@
-/* Theme, WebSocket, recording, rendering logic extracted from inline script and adapted for segmented theme control and WS caption */
+const isExtension = typeof chrome !== 'undefined' && chrome.runtime && chrome.runtime.getURL;
+if (isExtension) {
+  document.documentElement.classList.add('is-extension');
+}
+const isWebContext = !isExtension;
 
 let isRecording = false;
 let websocket = null;
@@ -12,6 +16,8 @@ let timerInterval = null;
 let audioContext = null;
 let analyser = null;
 let microphone = null;
+let workletNode = null;
+let recorderWorker = null;
 let waveCanvas = document.getElementById("waveCanvas");
 let waveCtx = waveCanvas.getContext("2d");
 let animationFrame = null;
@@ -20,6 +26,11 @@ let lastReceivedData = null;
 let lastSignature = null;
 let availableMicrophones = [];
 let selectedMicrophoneId = null;
+let serverUseAudioWorklet = null;
+let configReadyResolve;
+const configReady = new Promise((r) => (configReadyResolve = r));
+let outputAudioContext = null;
+let audioSource = null;
 
 waveCanvas.width = 60 * (window.devicePixelRatio || 1);
 waveCanvas.height = 30 * (window.devicePixelRatio || 1);
@@ -35,6 +46,26 @@ const timerElement = document.querySelector(".timer");
 const themeRadios = document.querySelectorAll('input[name="theme"]');
 const microphoneSelect = document.getElementById("microphoneSelect");
 
+const settingsToggle = document.getElementById("settingsToggle");
+const settingsDiv = document.querySelector(".settings");
+
+// if (isExtension) {
+//   chrome.runtime.onInstalled.addListener((details) => {
+//     if (details.reason.search(/install/g) === -1) {
+//       return;
+//     }
+//     chrome.tabs.create({
+//       url: chrome.runtime.getURL("welcome.html"),
+//       active: true
+//     });
+//   });
+// }
+
+const translationIcon = `<svg xmlns="http://www.w3.org/2000/svg" height="12px" viewBox="0 -960 960 960" width="12px" fill="#5f6368"><path d="m603-202-34 97q-4 11-14 18t-22 7q-20 0-32.5-16.5T496-133l152-402q5-11 15-18t22-7h30q12 0 22 7t15 18l152 403q8 19-4 35.5T868-80q-13 0-22.5-7T831-106l-34-96H603ZM362-401 188-228q-11 11-27.5 11.5T132-228q-11-11-11-28t11-28l174-174q-35-35-63.5-80T190-640h84q20 39 40 68t48 58q33-33 68.5-92.5T484-720H80q-17 0-28.5-11.5T40-760q0-17 11.5-28.5T80-800h240v-40q0-17 11.5-28.5T360-880q17 0 28.5 11.5T400-840v40h240q17 0 28.5 11.5T680-760q0 17-11.5 28.5T640-720h-76q-21 72-63 148t-83 116l96 98-30 82-122-125Zm266 129h144l-72-204-72 204Z"/></svg>`
+const silenceIcon = `<svg xmlns="http://www.w3.org/2000/svg" style="vertical-align: text-bottom;" height="14px" viewBox="0 -960 960 960" width="14px" fill="#5f6368"><path d="M514-556 320-752q9-3 19-5.5t21-2.5q66 0 113 47t47 113q0 11-1.5 22t-4.5 22ZM40-200v-32q0-33 17-62t47-44q51-26 115-44t141-18q26 0 49.5 2.5T456-392l-56-54q-9 3-19 4.5t-21 1.5q-66 0-113-47t-47-113q0-11 1.5-21t4.5-19L84-764q-11-11-11-28t11-28q12-12 28.5-12t27.5 12l675 685q11 11 11.5 27.5T816-80q-11 13-28 12.5T759-80L641-200h39q0 33-23.5 56.5T600-120H120q-33 0-56.5-23.5T40-200Zm80 0h480v-32q0-14-4.5-19.5T580-266q-36-18-92.5-36T360-320q-71 0-127.5 18T140-266q-9 5-14.5 14t-5.5 20v32Zm240 0Zm560-400q0 69-24.5 131.5T829-355q-12 14-30 15t-32-13q-13-13-12-31t12-33q30-38 46.5-85t16.5-98q0-51-16.5-97T767-781q-12-15-12.5-33t12.5-32q13-14 31.5-13.5T829-845q42 51 66.5 113.5T920-600Zm-182 0q0 32-10 61.5T700-484q-11 15-29.5 15.5T638-482q-13-13-13.5-31.5T633-549q6-11 9.5-24t3.5-27q0-14-3.5-27t-9.5-25q-9-17-8.5-35t13.5-31q14-14 32.5-13.5T700-716q18 25 28 54.5t10 61.5Z"/></svg>`;
+const languageIcon = `<svg xmlns="http://www.w3.org/2000/svg" height="12" viewBox="0 -960 960 960" width="12" fill="#5f6368"><path d="M480-80q-82 0-155-31.5t-127.5-86Q143-252 111.5-325T80-480q0-83 31.5-155.5t86-127Q252-817 325-848.5T480-880q83 0 155.5 31.5t127 86q54.5 54.5 86 127T880-480q0 82-31.5 155t-86 127.5q-54.5 54.5-127 86T480-80Zm0-82q26-36 45-75t31-83H404q12 44 31 83t45 75Zm-104-16q-18-33-31.5-68.5T322-320H204q29 50 72.5 87t99.5 55Zm208 0q56-18 99.5-55t72.5-87H638q-9 38-22.5 73.5T584-178ZM170-400h136q-3-20-4.5-39.5T300-480q0-21 1.5-40.5T306-560H170q-5 20-7.5 39.5T160-480q0 21 2.5 40.5T170-400Zm216 0h188q3-20 4.5-39.5T580-480q0-21-1.5-40.5T574-560H386q-3 20-4.5 39.5T380-480q0 21 1.5 40.5T386-400Zm268 0h136q5-20 7.5-39.5T800-480q0-21-2.5-40.5T790-560H654q3 20 4.5 39.5T660-480q0 21-1.5 40.5T654-400Zm-16-240h118q-29-50-72.5-87T584-782q18 33 31.5 68.5T638-640Zm-234 0h152q-12-44-31-83t-45-75q-26 36-45 75t-31 83Zm-200 0h118q9-38 22.5-73.5T376-782q-56 18-99.5 55T204-640Z"/></svg>`
+const speakerIcon = `<svg xmlns="http://www.w3.org/2000/svg" height="16px" style="vertical-align: text-bottom;" viewBox="0 -960 960 960" width="16px" fill="#5f6368"><path d="M480-480q-66 0-113-47t-47-113q0-66 47-113t113-47q66 0 113 47t47 113q0 66-47 113t-113 47ZM160-240v-32q0-34 17.5-62.5T224-378q62-31 126-46.5T480-440q66 0 130 15.5T736-378q29 15 46.5 43.5T800-272v32q0 33-23.5 56.5T720-160H240q-33 0-56.5-23.5T160-240Zm80 0h480v-32q0-11-5.5-20T700-306q-54-27-109-40.5T480-360q-56 0-111 13.5T260-306q-9 5-14.5 14t-5.5 20v32Zm240-320q33 0 56.5-23.5T560-640q0-33-23.5-56.5T480-720q-33 0-56.5 23.5T400-640q0 33 23.5 56.5T480-560Zm0-80Zm0 400Z"/></svg>`;
+
 function getWaveStroke() {
   const styles = getComputedStyle(document.documentElement);
   const v = styles.getPropertyValue("--wave-stroke").trim();
@@ -146,10 +177,16 @@ function fmt1(x) {
   return Number.isFinite(n) ? n.toFixed(1) : x;
 }
 
-// Default WebSocket URL computation
-const host = window.location.hostname || "localhost";
-const port = window.location.port;
-const protocol = window.location.protocol === "https:" ? "wss" : "ws";
+let host, port, protocol;
+port = 8000;
+if (isExtension) {
+    host = "localhost";
+    protocol = "ws";
+} else {
+    host = window.location.hostname || "localhost";
+    port = window.location.port;
+    protocol = window.location.protocol === "https:" ? "wss" : "ws";
+}
 const defaultWebSocketUrl = `${protocol}://${host}${port ? ":" + port : ""}/asr`;
 
 // Populate default caption and input
@@ -226,6 +263,14 @@ function setupWebSocket() {
 
     websocket.onmessage = (event) => {
       const data = JSON.parse(event.data);
+      if (data.type === "config") {
+        serverUseAudioWorklet = !!data.useAudioWorklet;
+        statusText.textContent = serverUseAudioWorklet
+          ? "Connected. Using AudioWorklet (PCM)."
+          : "Connected. Using MediaRecorder (WebM).";
+        if (configReadyResolve) configReadyResolve();
+        return;
+      }
 
       if (data.type === "ready_to_stop") {
         console.log("Ready to stop received, finalizing display and closing WebSocket.");
@@ -293,7 +338,7 @@ function renderLinesWithBuffer(
   const showTransLag = !isFinalizing && remaining_time_transcription > 0;
   const showDiaLag = !isFinalizing && !!buffer_diarization && remaining_time_diarization > 0;
   const signature = JSON.stringify({
-    lines: (lines || []).map((it) => ({ speaker: it.speaker, text: it.text, start: it.start, end: it.end })),
+    lines: (lines || []).map((it) => ({ speaker: it.speaker, text: it.text, start: it.start, end: it.end, detected_language: it.detected_language })),
     buffer_transcription: buffer_transcription || "",
     buffer_diarization: buffer_diarization || "",
     status: current_status,
@@ -322,23 +367,21 @@ function renderLinesWithBuffer(
 
       let speakerLabel = "";
       if (item.speaker === -2) {
-        speakerLabel = `<span class="silence">Silence<span id='timeInfo'>${timeInfo}</span></span>`;
+        speakerLabel = `<span class="silence">${silenceIcon}<span id='timeInfo'>${timeInfo}</span></span>`;
       } else if (item.speaker == 0 && !isFinalizing) {
         speakerLabel = `<span class='loading'><span class="spinner"></span><span id='timeInfo'><span class="loading-diarization-value">${fmt1(
           remaining_time_diarization
         )}</span> second(s) of audio are undergoing diarization</span></span>`;
       } else if (item.speaker !== 0) {
-        speakerLabel = `<span id="speaker">Speaker ${item.speaker}<span id='timeInfo'>${timeInfo}</span></span>`;
+        const speakerNum = `<span class="speaker-badge">${item.speaker}</span>`;
+        speakerLabel = `<span id="speaker">${speakerIcon}${speakerNum}<span id='timeInfo'>${timeInfo}</span></span>`;
+
+        if (item.detected_language) {
+          speakerLabel += `<span class="label_language">${languageIcon}<span>${item.detected_language}</span></span>`;
+        }
       }
 
       let currentLineText = item.text || "";
-      
-      if (item.translation) {
-        currentLineText += `<div class="label_translation">
-          <img src="/web/src/translate.svg" alt="Translation" width="12" height="12" />
-          <span>${item.translation}</span>
-        </div>`;
-      }
 
       if (idx === lines.length - 1) {
         if (!isFinalizing && item.speaker !== -2) {
@@ -372,6 +415,16 @@ function renderLinesWithBuffer(
           }
         }
       }
+      
+      if (item.translation) {
+        currentLineText += `
+            <div>
+                <div class="label_translation">
+                    ${translationIcon}
+                    <span>${item.translation}</span>
+                </div>
+            </div>`;
+      }
 
       return currentLineText.trim().length > 0 || speakerLabel.length > 0
         ? `<p>${speakerLabel}<br/><div class='textcontent'>${currentLineText}</div></p>`
@@ -445,11 +498,44 @@ async function startRecording() {
       console.log("Error acquiring wake lock.");
     }
 
-    const audioConstraints = selectedMicrophoneId 
-      ? { audio: { deviceId: { exact: selectedMicrophoneId } } }
-      : { audio: true };
-
-    const stream = await navigator.mediaDevices.getUserMedia(audioConstraints);
+    let stream;
+    
+    // chromium extension. in the future, both chrome page audio and mic will be used
+    if (isExtension) {
+      try {
+        stream = await new Promise((resolve, reject) => {
+          chrome.tabCapture.capture({audio: true}, (s) => {
+            if (s) {
+              resolve(s);
+            } else {
+              reject(new Error('Tab capture failed or not available'));
+            }
+          });
+        });
+        
+        try {
+          outputAudioContext = new (window.AudioContext || window.webkitAudioContext)();
+          audioSource = outputAudioContext.createMediaStreamSource(stream);
+          audioSource.connect(outputAudioContext.destination);
+        } catch (audioError) {
+          console.warn('could not preserve system audio:', audioError);
+        }
+        
+        statusText.textContent = "Using tab audio capture.";
+      } catch (tabError) {
+        console.log('Tab capture not available, falling back to microphone', tabError);
+        const audioConstraints = selectedMicrophoneId
+          ? { audio: { deviceId: { exact: selectedMicrophoneId } } }
+          : { audio: true };
+        stream = await navigator.mediaDevices.getUserMedia(audioConstraints);
+        statusText.textContent = "Using microphone audio.";
+      }
+    } else if (isWebContext) {
+      const audioConstraints = selectedMicrophoneId 
+        ? { audio: { deviceId: { exact: selectedMicrophoneId } } }
+        : { audio: true };
+      stream = await navigator.mediaDevices.getUserMedia(audioConstraints);
+    }
 
     audioContext = new (window.AudioContext || window.webkitAudioContext)();
     analyser = audioContext.createAnalyser();
@@ -457,13 +543,54 @@ async function startRecording() {
     microphone = audioContext.createMediaStreamSource(stream);
     microphone.connect(analyser);
 
-    recorder = new MediaRecorder(stream, { mimeType: "audio/webm" });
-    recorder.ondataavailable = (e) => {
-      if (websocket && websocket.readyState === WebSocket.OPEN) {
-        websocket.send(e.data);
+    if (serverUseAudioWorklet) {
+      if (!audioContext.audioWorklet) {
+        throw new Error("AudioWorklet is not supported in this browser");
       }
-    };
-    recorder.start(chunkDuration);
+      await audioContext.audioWorklet.addModule("/web/pcm_worklet.js");
+      workletNode = new AudioWorkletNode(audioContext, "pcm-forwarder", { numberOfInputs: 1, numberOfOutputs: 0, channelCount: 1 });
+      microphone.connect(workletNode);
+
+      recorderWorker = new Worker("/web/recorder_worker.js");
+      recorderWorker.postMessage({
+        command: "init",
+        config: {
+          sampleRate: audioContext.sampleRate,
+        },
+      });
+
+      recorderWorker.onmessage = (e) => {
+        if (websocket && websocket.readyState === WebSocket.OPEN) {
+          websocket.send(e.data.buffer);
+        }
+      };
+
+      workletNode.port.onmessage = (e) => {
+        const data = e.data;
+        const ab = data instanceof ArrayBuffer ? data : data.buffer;
+        recorderWorker.postMessage(
+          {
+            command: "record",
+            buffer: ab,
+          },
+          [ab]
+        );
+      };
+    } else {
+      try {
+        recorder = new MediaRecorder(stream, { mimeType: "audio/webm" });
+      } catch (e) {
+        recorder = new MediaRecorder(stream);
+      }
+      recorder.ondataavailable = (e) => {
+        if (websocket && websocket.readyState === WebSocket.OPEN) {
+          if (e.data && e.data.size > 0) {
+            websocket.send(e.data);
+          }
+        }
+      };
+      recorder.start(chunkDuration);
+    }
 
     startTime = Date.now();
     timerInterval = setInterval(updateTimer, 1000);
@@ -502,10 +629,28 @@ async function stopRecording() {
   }
 
   if (recorder) {
-    recorder.stop();
+    try {
+      recorder.stop();
+    } catch (e) {
+    }
     recorder = null;
   }
 
+  if (recorderWorker) {
+    recorderWorker.terminate();
+    recorderWorker = null;
+  }
+  
+  if (workletNode) {
+    try {
+      workletNode.port.onmessage = null;
+    } catch (e) {}
+    try {
+      workletNode.disconnect();
+    } catch (e) {}
+    workletNode = null;
+  }
+
   if (microphone) {
     microphone.disconnect();
     microphone = null;
@@ -524,6 +669,16 @@ async function stopRecording() {
     audioContext = null;
   }
 
+  if (audioSource) {
+    audioSource.disconnect();
+    audioSource = null;
+  }
+
+  if (outputAudioContext && outputAudioContext.state !== "closed") {
+    outputAudioContext.close()
+    outputAudioContext = null;
+  }
+
   if (animationFrame) {
     cancelAnimationFrame(animationFrame);
     animationFrame = null;
@@ -549,9 +704,11 @@ async function toggleRecording() {
     console.log("Connecting to WebSocket");
     try {
       if (websocket && websocket.readyState === WebSocket.OPEN) {
+        await configReady;
         await startRecording();
       } else {
         await setupWebSocket();
+        await configReady;
         await startRecording();
       }
     } catch (err) {
@@ -573,7 +730,7 @@ function updateUI() {
       statusText.textContent = "Please wait for processing to complete...";
     }
   } else if (isRecording) {
-    statusText.textContent = "Recording...";
+    statusText.textContent = "";
   } else {
     if (
       statusText.textContent !== "Finished processing audio! Ready to record again." &&
@@ -607,3 +764,40 @@ navigator.mediaDevices.addEventListener('devicechange', async () => {
     console.log("Error re-enumerating microphones:", error);
   }
 });
+
+
+settingsToggle.addEventListener("click", () => {
+settingsDiv.classList.toggle("visible");
+settingsToggle.classList.toggle("active");
+});
+
+if (isExtension) {
+  async function checkAndRequestPermissions() {
+    const micPermission = await navigator.permissions.query({
+      name: "microphone",
+    });
+
+    const permissionDisplay = document.getElementById("audioPermission");
+    if (permissionDisplay) {
+      permissionDisplay.innerText = `MICROPHONE: ${micPermission.state}`;
+    }
+
+    // if (micPermission.state !== "granted") {
+    //   chrome.tabs.create({ url: "welcome.html" });
+    // }
+
+    const intervalId = setInterval(async () => {
+      const micPermission = await navigator.permissions.query({
+        name: "microphone",
+      });
+      if (micPermission.state === "granted") {
+        if (permissionDisplay) {
+          permissionDisplay.innerText = `MICROPHONE: ${micPermission.state}`;
+        }
+        clearInterval(intervalId);
+      }
+    }, 100);
+  }
+
+  void checkAndRequestPermissions();
+}

whisperlivekit/web/pcm_worklet.js

@@ -0,0 +1,16 @@
+class PCMForwarder extends AudioWorkletProcessor {
+  process(inputs) {
+    const input = inputs[0];
+    if (input && input[0] && input[0].length) {
+      // Forward mono channel (0). If multi-channel, downmixing can be added here.
+      const channelData = input[0];
+      const copy = new Float32Array(channelData.length);
+      copy.set(channelData);
+      this.port.postMessage(copy, [copy.buffer]);
+    }
+    // Keep processor alive
+    return true;
+  }
+}
+
+registerProcessor('pcm-forwarder', PCMForwarder);

whisperlivekit/web/recorder_worker.js

@@ -0,0 +1,58 @@
+let sampleRate = 48000;
+let targetSampleRate = 16000;
+
+self.onmessage = function (e) {
+  switch (e.data.command) {
+    case 'init':
+      init(e.data.config);
+      break;
+    case 'record':
+      record(e.data.buffer);
+      break;
+  }
+};
+
+function init(config) {
+  sampleRate = config.sampleRate;
+  targetSampleRate = config.targetSampleRate || 16000;
+}
+
+function record(inputBuffer) {
+  const buffer = new Float32Array(inputBuffer);
+  const resampledBuffer = resample(buffer, sampleRate, targetSampleRate);
+  const pcmBuffer = toPCM(resampledBuffer);
+  self.postMessage({ buffer: pcmBuffer }, [pcmBuffer]);
+}
+
+function resample(buffer, from, to) {
+    if (from === to) {
+        return buffer;
+    }
+    const ratio = from / to;
+    const newLength = Math.round(buffer.length / ratio);
+    const result = new Float32Array(newLength);
+    let offsetResult = 0;
+    let offsetBuffer = 0;
+    while (offsetResult < result.length) {
+        const nextOffsetBuffer = Math.round((offsetResult + 1) * ratio);
+        let accum = 0, count = 0;
+        for (let i = offsetBuffer; i < nextOffsetBuffer && i < buffer.length; i++) {
+            accum += buffer[i];
+            count++;
+        }
+        result[offsetResult] = accum / count;
+        offsetResult++;
+        offsetBuffer = nextOffsetBuffer;
+    }
+    return result;
+}
+
+function toPCM(input) {
+  const buffer = new ArrayBuffer(input.length * 2);
+  const view = new DataView(buffer);
+  for (let i = 0; i < input.length; i++) {
+    const s = Math.max(-1, Math.min(1, input[i]));
+    view.setInt16(i * 2, s < 0 ? s * 0x8000 : s * 0x7FFF, true);
+  }
+  return buffer;
+}

whisperlivekit/web/src/language.svg

@@ -0,0 +1 @@
+<svg xmlns="http://www.w3.org/2000/svg" height="24px" viewBox="0 -960 960 960" width="24px" fill="#5f6368"><path d="M480-80q-82 0-155-31.5t-127.5-86Q143-252 111.5-325T80-480q0-83 31.5-155.5t86-127Q252-817 325-848.5T480-880q83 0 155.5 31.5t127 86q54.5 54.5 86 127T880-480q0 82-31.5 155t-86 127.5q-54.5 54.5-127 86T480-80Zm0-82q26-36 45-75t31-83H404q12 44 31 83t45 75Zm-104-16q-18-33-31.5-68.5T322-320H204q29 50 72.5 87t99.5 55Zm208 0q56-18 99.5-55t72.5-87H638q-9 38-22.5 73.5T584-178ZM170-400h136q-3-20-4.5-39.5T300-480q0-21 1.5-40.5T306-560H170q-5 20-7.5 39.5T160-480q0 21 2.5 40.5T170-400Zm216 0h188q3-20 4.5-39.5T580-480q0-21-1.5-40.5T574-560H386q-3 20-4.5 39.5T380-480q0 21 1.5 40.5T386-400Zm268 0h136q5-20 7.5-39.5T800-480q0-21-2.5-40.5T790-560H654q3 20 4.5 39.5T660-480q0 21-1.5 40.5T654-400Zm-16-240h118q-29-50-72.5-87T584-782q18 33 31.5 68.5T638-640Zm-234 0h152q-12-44-31-83t-45-75q-26 36-45 75t-31 83Zm-200 0h118q9-38 22.5-73.5T376-782q-56 18-99.5 55T204-640Z"/></svg>

whisperlivekit/web/src/silence.svg

@@ -0,0 +1 @@
+<svg xmlns="http://www.w3.org/2000/svg" height="24px" viewBox="0 -960 960 960" width="24px" fill="#5f6368"><path d="M514-556 320-752q9-3 19-5.5t21-2.5q66 0 113 47t47 113q0 11-1.5 22t-4.5 22ZM40-200v-32q0-33 17-62t47-44q51-26 115-44t141-18q26 0 49.5 2.5T456-392l-56-54q-9 3-19 4.5t-21 1.5q-66 0-113-47t-47-113q0-11 1.5-21t4.5-19L84-764q-11-11-11-28t11-28q12-12 28.5-12t27.5 12l675 685q11 11 11.5 27.5T816-80q-11 13-28 12.5T759-80L641-200h39q0 33-23.5 56.5T600-120H120q-33 0-56.5-23.5T40-200Zm80 0h480v-32q0-14-4.5-19.5T580-266q-36-18-92.5-36T360-320q-71 0-127.5 18T140-266q-9 5-14.5 14t-5.5 20v32Zm240 0Zm560-400q0 69-24.5 131.5T829-355q-12 14-30 15t-32-13q-13-13-12-31t12-33q30-38 46.5-85t16.5-98q0-51-16.5-97T767-781q-12-15-12.5-33t12.5-32q13-14 31.5-13.5T829-845q42 51 66.5 113.5T920-600Zm-182 0q0 32-10 61.5T700-484q-11 15-29.5 15.5T638-482q-13-13-13.5-31.5T633-549q6-11 9.5-24t3.5-27q0-14-3.5-27t-9.5-25q-9-17-8.5-35t13.5-31q14-14 32.5-13.5T700-716q18 25 28 54.5t10 61.5Z"/></svg>

whisperlivekit/web/src/speaker.svg

@@ -0,0 +1 @@
+<svg xmlns="http://www.w3.org/2000/svg" height="24px" viewBox="0 -960 960 960" width="24px" fill="#5f6368"><path d="M480-480q-66 0-113-47t-47-113q0-66 47-113t113-47q66 0 113 47t47 113q0 66-47 113t-113 47ZM160-240v-32q0-34 17.5-62.5T224-378q62-31 126-46.5T480-440q66 0 130 15.5T736-378q29 15 46.5 43.5T800-272v32q0 33-23.5 56.5T720-160H240q-33 0-56.5-23.5T160-240Zm80 0h480v-32q0-11-5.5-20T700-306q-54-27-109-40.5T480-360q-56 0-111 13.5T260-306q-9 5-14.5 14t-5.5 20v32Zm240-320q33 0 56.5-23.5T560-640q0-33-23.5-56.5T480-720q-33 0-56.5 23.5T400-640q0 33 23.5 56.5T480-560Zm0-80Zm0 400Z"/></svg>

whisperlivekit/web/web_interface.py

@@ -23,6 +23,24 @@ def get_inline_ui_html():
         with resources.files('whisperlivekit.web').joinpath('live_transcription.js').open('r', encoding='utf-8') as f:
             js_content = f.read()
         
+        with resources.files('whisperlivekit.web').joinpath('pcm_worklet.js').open('r', encoding='utf-8') as f:
+            worklet_code = f.read()
+        with resources.files('whisperlivekit.web').joinpath('recorder_worker.js').open('r', encoding='utf-8') as f:
+            worker_code = f.read()
+        
+        js_content = js_content.replace(
+            'await audioContext.audioWorklet.addModule("/web/pcm_worklet.js");',
+            'const workletBlob = new Blob([`' + worklet_code + '`], { type: "application/javascript" });\n' +
+            'const workletUrl = URL.createObjectURL(workletBlob);\n' +
+            'await audioContext.audioWorklet.addModule(workletUrl);'
+        )
+        js_content = js_content.replace(
+            'recorderWorker = new Worker("/web/recorder_worker.js");',
+            'const workerBlob = new Blob([`' + worker_code + '`], { type: "application/javascript" });\n' +
+            'const workerUrl = URL.createObjectURL(workerBlob);\n' +
+            'recorderWorker = new Worker(workerUrl);'
+        )
+        
         # SVG files
         with resources.files('whisperlivekit.web').joinpath('src', 'system_mode.svg').open('r', encoding='utf-8') as f:
             system_svg = f.read()
@@ -33,15 +51,18 @@ def get_inline_ui_html():
         with resources.files('whisperlivekit.web').joinpath('src', 'dark_mode.svg').open('r', encoding='utf-8') as f:
             dark_svg = f.read()
             dark_data_uri = f"data:image/svg+xml;base64,{base64.b64encode(dark_svg.encode('utf-8')).decode('utf-8')}"
-        
+        with resources.files('whisperlivekit.web').joinpath('src', 'settings.svg').open('r', encoding='utf-8') as f:
+            settings = f.read()
+            settings_uri = f"data:image/svg+xml;base64,{base64.b64encode(settings.encode('utf-8')).decode('utf-8')}"
+
         # Replace external references
         html_content = html_content.replace(
-            '<link rel="stylesheet" href="/web/live_transcription.css" />',
+            '<link rel="stylesheet" href="live_transcription.css" />',
             f'<style>\n{css_content}\n</style>'
         )
         
         html_content = html_content.replace(
-            '<script src="/web/live_transcription.js"></script>',
+            '<script src="live_transcription.js"></script>',
             f'<script>\n{js_content}\n</script>'
         )
         
@@ -61,6 +82,11 @@ def get_inline_ui_html():
             f'<img src="{dark_data_uri}" alt="" />'
         )
         
+        html_content = html_content.replace(
+            '<img src="web/src/settings.svg" alt="Settings" />',
+            f'<img src="{settings_uri}" alt="" />'
+        )
+        
         return html_content
         
     except Exception as e:

whisperlivekit/whisper_streaming_custom/backends.py

@@ -11,14 +11,14 @@ class ASRBase:
     sep = " "  # join transcribe words with this character (" " for whisper_timestamped,
               # "" for faster-whisper because it emits the spaces when needed)
 
-    def __init__(self, lan, modelsize=None, cache_dir=None, model_dir=None, logfile=sys.stderr):
+    def __init__(self, lan, model_size=None, cache_dir=None, model_dir=None, logfile=sys.stderr):
         self.logfile = logfile
         self.transcribe_kargs = {}
         if lan == "auto":
             self.original_language = None
         else:
             self.original_language = lan
-        self.model = self.load_model(modelsize, cache_dir, model_dir)
+        self.model = self.load_model(model_size, cache_dir, model_dir)
 
     def with_offset(self, offset: float) -> ASRToken:
         # This method is kept for compatibility (typically you will use ASRToken.with_offset)
@@ -27,7 +27,7 @@ class ASRBase:
     def __repr__(self):
         return f"ASRToken(start={self.start:.2f}, end={self.end:.2f}, text={self.text!r})"
 
-    def load_model(self, modelsize, cache_dir, model_dir):
+    def load_model(self, model_size, cache_dir, model_dir):
         raise NotImplementedError("must be implemented in the child class")
 
     def transcribe(self, audio, init_prompt=""):
@@ -41,7 +41,7 @@ class WhisperTimestampedASR(ASRBase):
     """Uses whisper_timestamped as the backend."""
     sep = " "
 
-    def load_model(self, modelsize=None, cache_dir=None, model_dir=None):
+    def load_model(self, model_size=None, cache_dir=None, model_dir=None):
         import whisper
         import whisper_timestamped
         from whisper_timestamped import transcribe_timestamped
@@ -49,7 +49,7 @@ class WhisperTimestampedASR(ASRBase):
         self.transcribe_timestamped = transcribe_timestamped
         if model_dir is not None:
             logger.debug("ignoring model_dir, not implemented")
-        return whisper.load_model(modelsize, download_root=cache_dir)
+        return whisper.load_model(model_size, download_root=cache_dir)
 
     def transcribe(self, audio, init_prompt=""):
         result = self.transcribe_timestamped(
@@ -88,17 +88,17 @@ class FasterWhisperASR(ASRBase):
     """Uses faster-whisper as the backend."""
     sep = ""
 
-    def load_model(self, modelsize=None, cache_dir=None, model_dir=None):
+    def load_model(self, model_size=None, cache_dir=None, model_dir=None):
         from faster_whisper import WhisperModel
 
         if model_dir is not None:
             logger.debug(f"Loading whisper model from model_dir {model_dir}. "
-                         f"modelsize and cache_dir parameters are not used.")
+                         f"model_size and cache_dir parameters are not used.")
             model_size_or_path = model_dir
-        elif modelsize is not None:
-            model_size_or_path = modelsize
+        elif model_size is not None:
+            model_size_or_path = model_size
         else:
-            raise ValueError("Either modelsize or model_dir must be set")
+            raise ValueError("Either model_size or model_dir must be set")
         device = "auto" # Allow CTranslate2 to decide available device
         compute_type = "auto" # Allow CTranslate2 to decide faster compute type
                               
@@ -149,18 +149,18 @@ class MLXWhisper(ASRBase):
     """
     sep = ""
 
-    def load_model(self, modelsize=None, cache_dir=None, model_dir=None):
+    def load_model(self, model_size=None, cache_dir=None, model_dir=None):
         from mlx_whisper.transcribe import ModelHolder, transcribe
         import mlx.core as mx
 
         if model_dir is not None:
-            logger.debug(f"Loading whisper model from model_dir {model_dir}. modelsize parameter is not used.")
+            logger.debug(f"Loading whisper model from model_dir {model_dir}. model_size parameter is not used.")
             model_size_or_path = model_dir
-        elif modelsize is not None:
-            model_size_or_path = self.translate_model_name(modelsize)
-            logger.debug(f"Loading whisper model {modelsize}. You use mlx whisper, so {model_size_or_path} will be used.")
+        elif model_size is not None:
+            model_size_or_path = self.translate_model_name(model_size)
+            logger.debug(f"Loading whisper model {model_size}. You use mlx whisper, so {model_size_or_path} will be used.")
         else:
-            raise ValueError("Either modelsize or model_dir must be set")
+            raise ValueError("Either model_size or model_dir must be set")
 
         self.model_size_or_path = model_size_or_path
         dtype = mx.float16

whisperlivekit/whisper_streaming_custom/online_asr.py

@@ -106,9 +106,6 @@ class OnlineASRProcessor:
     def __init__(
         self,
         asr,
-        tokenize_method: Optional[callable] = None,
-        buffer_trimming: Tuple[str, float] = ("segment", 15),
-        confidence_validation = False,
         logfile=sys.stderr,
     ):
         """
@@ -119,13 +116,14 @@ class OnlineASRProcessor:
         buffer_trimming: A tuple (option, seconds), where option is either "sentence" or "segment".
         """
         self.asr = asr
-        self.tokenize = tokenize_method
+        self.tokenize = asr.tokenizer
         self.logfile = logfile
-        self.confidence_validation = confidence_validation
+        self.confidence_validation = asr.confidence_validation
         self.global_time_offset = 0.0
         self.init()
 
-        self.buffer_trimming_way, self.buffer_trimming_sec = buffer_trimming
+        self.buffer_trimming_way = asr.buffer_trimming
+        self.buffer_trimming_sec = asr.buffer_trimming_sec
 
         if self.buffer_trimming_way not in ["sentence", "segment"]:
             raise ValueError("buffer_trimming must be either 'sentence' or 'segment'")

whisperlivekit/whisper_streaming_custom/whisper_online.py

@@ -6,6 +6,7 @@ from functools import lru_cache
 import time
 import logging
 from .backends import FasterWhisperASR, MLXWhisper, WhisperTimestampedASR, OpenaiApiASR
+from whisperlivekit.warmup import warmup_asr
 
 logger = logging.getLogger(__name__)
 
@@ -63,11 +64,23 @@ def create_tokenizer(lan):
     return WtPtok()
 
 
-def backend_factory(args):
-    backend = args.backend
+def backend_factory(
+            backend,
+            lan,
+            model_size,
+            model_cache_dir,
+            model_dir,
+            task,
+            buffer_trimming,
+            buffer_trimming_sec,
+            confidence_validation,
+            warmup_file=None,
+            min_chunk_size=None,
+        ):
+    backend = backend
     if backend == "openai-api":
         logger.debug("Using OpenAI API.")
-        asr = OpenaiApiASR(lan=args.lan)        
+        asr = OpenaiApiASR(lan=lan)        
     else:
         if backend == "faster-whisper":
             asr_cls = FasterWhisperASR
@@ -77,34 +90,33 @@ def backend_factory(args):
             asr_cls = WhisperTimestampedASR
 
         # Only for FasterWhisperASR and WhisperTimestampedASR
-        size = args.model
+
         t = time.time()
-        logger.info(f"Loading Whisper {size} model for language {args.lan}...")
+        logger.info(f"Loading Whisper {model_size} model for language {lan}...")
         asr = asr_cls(
-            modelsize=size,
-            lan=args.lan,
-            cache_dir=getattr(args, 'model_cache_dir', None),
-            model_dir=getattr(args, 'model_dir', None),
+            model_size=model_size,
+            lan=lan,
+            cache_dir=model_cache_dir,
+            model_dir=model_dir,
         )
         e = time.time()
         logger.info(f"done. It took {round(e-t,2)} seconds.")
 
-    # Apply common configurations
-    if getattr(args, "vad", False):  # Checks if VAD argument is present and True
-        logger.info("Setting VAD filter")
-        asr.use_vad()
-
-    language = args.lan
-    if args.task == "translate":
-        if backend != "simulstreaming":
-            asr.set_translate_task()
+    if task == "translate":
         tgt_language = "en"  # Whisper translates into English
     else:
-        tgt_language = language  # Whisper transcribes in this language
+        tgt_language = lan  # Whisper transcribes in this language
 
     # Create the tokenizer
-    if args.buffer_trimming == "sentence":
+    if buffer_trimming == "sentence":
         tokenizer = create_tokenizer(tgt_language)
     else:
         tokenizer = None
-    return asr, tokenizer
+    
+    warmup_asr(asr, warmup_file)
+    
+    asr.confidence_validation = confidence_validation
+    asr.tokenizer = tokenizer
+    asr.buffer_trimming = buffer_trimming
+    asr.buffer_trimming_sec = buffer_trimming_sec
+    return asr