solve #100

README.md

@@ -1,6 +1,11 @@
 <h1 align="center">WhisperLiveKit</h1>
 <p align="center"><b>Real-time, Fully Local Whisper's Speech-to-Text and Speaker Diarization</b></p>
 
+<p align="center">
+<img alt="PyPI Version" src="https://img.shields.io/pypi/v/whisperlivekit?color=g">
+<img alt="PyPI Downloads" src="https://static.pepy.tech/personalized-badge/whisperlivekit">
+<img alt="Python Versions" src="https://img.shields.io/badge/python-3.9%20%7C%203.10%20%7C%203.11%20%7C%203.12-dark_green">
+</p>
 
 This project is based on [Whisper Streaming](https://github.com/ufal/whisper_streaming) and lets you transcribe audio directly from your browser. Simply launch the local server and grant microphone access. Everything runs locally on your machine ✨
 
@@ -34,13 +39,11 @@ pip install whisperlivekit
 
 ### From source
 
-1. **Clone the Repository**:
+```bash
-
+git clone https://github.com/QuentinFuxa/WhisperLiveKit
-   ```bash
+cd WhisperLiveKit
-   git clone https://github.com/QuentinFuxa/WhisperLiveKit
+pip install -e .
-   cd WhisperLiveKit
+```
-   pip install -e .
-   ```
 
 ### System Dependencies
 
@@ -69,9 +72,25 @@ pip install tokenize_uk  # If you work with Ukrainian text
 
 # If you want to use diarization
 pip install diart
+
+# Optional backends. Default is faster-whisper
+pip install whisperlivekit[whisper]           # Original Whisper backend
+pip install whisperlivekit[whisper-timestamped]  # Whisper with improved timestamps
+pip install whisperlivekit[mlx-whisper]       # Optimized for Apple Silicon
+pip install whisperlivekit[openai]            # OpenAI API backend
 ```
 
-Diart uses [pyannote.audio](https://github.com/pyannote/pyannote-audio) models from the _huggingface hub_. To use them, please follow the steps described [here](https://github.com/juanmc2005/diart?tab=readme-ov-file#get-access-to--pyannote-models).
+### Get access to 🎹 pyannote models
+
+By default, diart is based on [pyannote.audio](https://github.com/pyannote/pyannote-audio) models from the [huggingface](https://huggingface.co/) hub.
+In order to use them, please follow these steps:
+
+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 newest `pyannote/segmentation-3.0` model
+3) [Accept user conditions](https://huggingface.co/pyannote/embedding) for the `pyannote/embedding` model
+4) Install [huggingface-cli](https://huggingface.co/docs/huggingface_hub/quick-start#install-the-hub-library) and [log in](https://huggingface.co/docs/huggingface_hub/quick-start#login) with your user access token (or provide it manually in diart CLI or API).
+
+
 
 ## Usage
 
@@ -123,32 +142,25 @@ For a complete audio processing example, check [whisper_fastapi_online_server.py
 The following parameters are supported when initializing `WhisperLiveKit`:
 
   - `--host` and `--port` let you specify the server's IP/port. 
-  - `-min-chunk-size` sets the minimum chunk size for audio processing. Make sure this value aligns with the chunk size selected in the frontend. If not aligned, the system will work but may unnecessarily over-process audio data.
+  - `--min-chunk-size` sets the minimum chunk size for audio processing. Make sure this value aligns with the chunk size selected in the frontend. If not aligned, the system will work but may unnecessarily over-process audio data.
-  - `--transcription`: Enable/disable transcription (default: True)
+  - `--no-transcription`: Disable transcription (enabled by default)
-  - `--diarization`: Enable/disable speaker diarization (default: False)
+  - `--diarization`: Enable speaker diarization (disabled by default)
-  - `--confidence-validation`: Use confidence scores for faster validation. Transcription will be faster but punctuation might be less accurate (default: True)
+  - `--confidence-validation`: Use confidence scores for faster validation. Transcription will be faster but punctuation might be less accurate (disabled by default)
-  - `--warmup-file`: The path to a speech audio wav file to warm up Whisper so that the very first chunk processing is fast. :
+  - `--warmup-file`: 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.
   - `--min-chunk-size` Minimum audio chunk size in seconds. It waits up to this time to do processing. If the processing takes shorter time, it waits, otherwise it processes the whole segment that was received by this time.
-  - `--model` {_tiny.en, tiny, base.en, base, small.en, small, medium.en, medium, large-v1, large-v2, large-v3, large, large-v3-turbo_}
+  - `--model`: 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.
-                        Name size of the Whisper model to use (default: tiny). The model is automatically downloaded from the model hub if not present in model cache dir.
+  - `--model_cache_dir`: Overriding the default model cache dir where models downloaded from the hub are saved
-  - `--model_cache_dir` Overriding the default model cache dir where models downloaded from the hub are saved
+  - `--model_dir`: Dir where Whisper model.bin and other files are saved. This option overrides --model and --model_cache_dir parameter.
-  - `--model_dir` Dir where Whisper model.bin and other files are saved. This option overrides --model and --model_cache_dir parameter.
+  - `--lan`, `--language`: Source language code, e.g. en,de,cs, or 'auto' for language detection.
-  - `--lan`, --language Source language code, e.g. en,de,cs, or 'auto' for language detection.
+  - `--task` {_transcribe, translate_}: Transcribe or translate. If translate is set, we recommend avoiding the _large-v3-turbo_ backend, as it [performs significantly worse](https://github.com/QuentinFuxa/whisper_streaming_web/issues/40#issuecomment-2652816533) than other models for translation.
-  - `--task` {_transcribe, translate_} Transcribe or translate. If translate is set, we recommend avoiding the _large-v3-turbo_ backend, as it [performs significantly worse](https://github.com/QuentinFuxa/whisper_streaming_web/issues/40#issuecomment-2652816533) than other models for translation.
+  - `--backend` {_faster-whisper, whisper_timestamped, openai-api, mlx-whisper_}: Load only this backend for Whisper processing.
-  - `--backend` {_faster-whisper, whisper_timestamped, openai-api, mlx-whisper_} Load only this backend for Whisper processing.
+  - `--vac`: Use VAC = voice activity controller. Requires torch. (disabled by default)
-  - `--vac` Use VAC = voice activity controller. Requires torch.
+  - `--vac-chunk-size`: VAC sample size in seconds.
-  - `--vac-chunk-size` VAC sample size in seconds.
+  - `--no-vad`: Disable VAD (voice activity detection), which is enabled by default.
-  - `--vad` Use VAD = voice activity detection, with the default parameters.
+  - `--buffer_trimming` {_sentence, segment_}: Buffer trimming strategy -- trim completed sentences marked with punctuation mark and detected by sentence segmenter, or the completed segments returned by Whisper. Sentence segmenter must be installed for "sentence" option.
-  - `--buffer_trimming` {_sentence, segment_} Buffer trimming strategy -- trim completed sentences marked with punctuation mark and detected by sentence segmenter, or the completed segments returned by Whisper. Sentence segmenter must be installed for "sentence" option.
+  - `--buffer_trimming_sec`: Buffer trimming length threshold in seconds. If buffer length is longer, trimming sentence/segment is triggered.
-  - `--buffer_trimming_sec` Buffer trimming length threshold in seconds. If buffer length is longer, trimming sentence/segment is triggered.
-
-5. **Open the Provided HTML**:
-
-    - By default, the server root endpoint `/` serves a simple `live_transcription.html` page.  
-    - Open your browser at `http://localhost:8000` (or replace `localhost` and `8000` with whatever you specified).  
-    - The page uses vanilla JavaScript and the WebSocket API to capture your microphone and stream audio to the server in real time.
 
 
 ## How the Live Interface Works
@@ -157,7 +169,6 @@ The following parameters are supported when initializing `WhisperLiveKit`:
 - These chunks are sent over a **WebSocket** to the FastAPI endpoint at `/asr`.  
 - The Python server decodes `.webm` chunks on the fly using **FFmpeg** and streams them into the **whisper streaming** implementation for transcription.  
 - **Partial transcription** appears as soon as enough audio is processed. The "unvalidated" text is shown in **lighter or grey color** (i.e., an 'aperçu') to indicate it's still buffered partial output. Once Whisper finalizes that segment, it's displayed in normal text.
-- You can watch the transcription update in near real time, ideal for demos, prototyping, or quick debugging.
 
 ### Deploying to a Remote Server
 
@@ -165,10 +176,8 @@ If you want to **deploy** this setup:
 
 1. **Host the FastAPI app** behind a production-grade HTTP(S) server (like **Uvicorn + Nginx** or Docker). If you use HTTPS, use "wss" instead of "ws" in WebSocket URL.
 2. The **HTML/JS page** can be served by the same FastAPI app or a separate static host.  
-3. Users open the page in **Chrome/Firefox** (any modern browser that supports MediaRecorder + WebSocket).  
+3. Users open the page in **Chrome/Firefox** (any modern browser that supports MediaRecorder + WebSocket). No additional front-end libraries or frameworks are required.
-
-No additional front-end libraries or frameworks are required. The WebSocket logic in `live_transcription.html` is minimal enough to adapt for your own custom UI or embed in other pages.
 
 ## Acknowledgments
 
-This project builds upon the foundational work of the Whisper Streaming project. We extend our gratitude to the original authors for their contributions.
+This project builds upon the foundational work of the Whisper Streaming and Diart projects. We extend our gratitude to the original authors for their contributions.

setup.py

@@ -1,8 +1,7 @@
 from setuptools import setup, find_packages
-
 setup(
     name="whisperlivekit",
-    version="0.1.0",
+    version="0.1.3",
     description="Real-time, Fully Local Whisper's Speech-to-Text and Speaker Diarization",
     long_description=open("README.md", "r", encoding="utf-8").read(),
     long_description_content_type="text/markdown",
@@ -22,6 +21,10 @@ setup(
         "diarization": ["diart"],
         "vac": ["torch"],
         "sentence": ["mosestokenizer", "wtpsplit"],
+        "whisper": ["whisper"],
+        "whisper-timestamped": ["whisper-timestamped"],
+        "mlx-whisper": ["mlx-whisper"],
+        "openai": ["openai"],
     },
     package_data={
         'whisperlivekit': ['web/*.html'],

whisperlivekit/core.py

@@ -1,7 +1,7 @@
 try:
     from whisperlivekit.whisper_streaming_custom.whisper_online import backend_factory, warmup_asr
-except:
+except ImportError:
-    from whisper_streaming_custom.whisper_online import backend_factory, warmup_asr
+    from .whisper_streaming_custom.whisper_online import backend_factory, warmup_asr
 from argparse import Namespace, ArgumentParser
 
 def parse_args():
@@ -29,23 +29,21 @@ def parse_args():
 
     parser.add_argument(
         "--confidence-validation",
-        type=bool,
+        action="store_true",
-        default=False,
         help="Accelerates validation of tokens using confidence scores. Transcription will be faster but punctuation might be less accurate.",
     )
 
     parser.add_argument(
         "--diarization",
-        type=bool,
+        action="store_true",
-        default=True,
+        default=False,
-        help="Whether to enable speaker diarization.",
+        help="Enable speaker diarization.",
     )
 
     parser.add_argument(
-        "--transcription",
+        "--no-transcription",
-        type=bool,
+        action="store_true",
-        default=True,
+        help="Disable transcription to only see live diarization results.",
-        help="To disable to only see live diarization results.",
     )
     
     parser.add_argument(
@@ -54,15 +52,14 @@ def parse_args():
         default=0.5,
         help="Minimum audio chunk size in seconds. It waits up to this time to do processing. If the processing takes shorter time, it waits, otherwise it processes the whole segment that was received by this time.",
     )
+    
     parser.add_argument(
         "--model",
         type=str,
         default="tiny",
-        choices="tiny.en,tiny,base.en,base,small.en,small,medium.en,medium,large-v1,large-v2,large-v3,large,large-v3-turbo".split(
+        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.",
-            ","
-        ),
-        help="Name size of the Whisper model to use (default: large-v2). The model is automatically downloaded from the model hub if not present in model cache dir.",
     )
+    
     parser.add_argument(
         "--model_cache_dir",
         type=str,
@@ -105,12 +102,13 @@ def parse_args():
     parser.add_argument(
         "--vac-chunk-size", type=float, default=0.04, help="VAC sample size in seconds."
     )
+
     parser.add_argument(
-        "--vad",
+        "--no-vad",
         action="store_true",
-        default=True,
+        help="Disable VAD (voice activity detection).",
-        help="Use VAD = voice activity detection, with the default parameters.",
     )
+    
     parser.add_argument(
         "--buffer_trimming",
         type=str,
@@ -134,6 +132,12 @@ def parse_args():
     )
 
     args = parser.parse_args()
+    
+    args.transcription = not args.no_transcription
+    args.vad = not args.no_vad    
+    delattr(args, 'no_transcription')
+    delattr(args, 'no_vad')
+    
     return args
 
 class WhisperLiveKit:

whisperlivekit/whisper_streaming_custom/backends.py

@@ -3,7 +3,10 @@ import logging
 import io
 import soundfile as sf
 import math
-import torch
+try: 
+    import torch
+except ImportError: 
+    torch = None
 from typing import List
 import numpy as np
 from whisperlivekit.timed_objects import ASRToken
@@ -102,7 +105,7 @@ class FasterWhisperASR(ASRBase):
             model_size_or_path = modelsize
         else:
             raise ValueError("Either modelsize or model_dir must be set")
-        device = "cuda" if torch.cuda.is_available() else "cpu"
+        device = "cuda" if torch and torch.cuda.is_available() else "cpu"
         compute_type = "float16" if device == "cuda" else "float32"
 
         model = WhisperModel(