78 Commits

Author	SHA1	Message	Date
Quentin Fuxa	f884d1162d	warning when transcribe_kargs are used with MLX Whisper	2025-01-14 20:14:16 +01:00
Quentin Fuxa	aa0ba598f0	no online conflict when multiple users	2025-01-03 14:48:45 +01:00
silask	6a04ddbed2	only print translated text not timestamps	2024-12-30 21:53:33 +01:00
silask	aa4d2599cc	fix #7	2024-12-30 21:53:33 +01:00
silask	5fdb08edae	black formating	2024-12-30 21:53:33 +01:00
Quentin Fuxa	8dcebd9329	add translate_model_name function	2024-12-19 11:10:02 +01:00
Quentin Fuxa	87cab7c280	add whisper mlx backend	2024-12-19 10:47:46 +01:00
Dominik Macháček	e6648e4f46	fixed silero vad chunk size ... issues #141 #121 #142 #136 etc.	2024-11-28 18:13:49 +01:00
Dario Pellegrino	d48895c343	large-v3-turbo compatible	2024-11-22 00:16:22 +01:00
Dominik Macháček	8cfd8d85a3	Merge branch 'main' of github.com:promet99/whisper_streaming into promet99-main	2024-11-15 13:53:57 +01:00
Dominik Macháček	e1b0e146a5	lru_cache didn't work with Python 3.6.9, openai api needs py version	2024-11-15 13:53:01 +01:00
promet99	e3dc524783	fix: update openapi code to match updated return type	2024-11-10 20:02:25 +09:00
Dominik Macháček	2de090023c	fixed VADIterator	2024-10-16 11:14:37 +02:00
Dominik Macháček	e25ad4fcd7	fixed ... issue #116	2024-10-04 17:39:20 +02:00
Dominik Macháček	63870987c0	FixedSileroVADIterator to support other than 512-sized chunks with v5 ... isssue #116	2024-10-04 17:14:55 +02:00
slaesh	d665f9a96e	fix: pin silero-vad version to v4	2024-09-01 08:05:40 +02:00
Dominik Macháček	827425bb91	small code review	2024-08-19 10:43:29 +02:00
Dominik Macháček	52da12120c	cleaner code	2024-08-19 00:19:16 +02:00
Dominik Macháček	7edc534f8a	clean code with VAC	2024-08-19 00:04:03 +02:00
Dominik Macháček	14c2bbef87	removing duplicated code -- whisper_online_vac	2024-08-18 20:33:08 +02:00
Dominik Macháček	36bf3a32d4	Merge branch 'main' into vad-streaming-clean ... conflicts merged	2024-08-18 19:47:28 +02:00
Ondrej Platek	13fd21a201	polishing code and note about installing deps for VAD	2024-08-16 16:53:07 +02:00
wenli	7bca7a2b8e	Update whisper_online.py	2024-05-15 23:49:24 +08:00
Dominik Macháček	264b8a32c2	forgot f in the last debug print	2024-05-02 10:32:53 +02:00
Dominik Macháček	b50f68749b	checks and changes in logging ... - don't set the level for submodules, it's too verbose - etc.	2024-04-18 18:10:25 +02:00
Alex Young	8060d45aea	Default log level to DEBUG, faster-whisper to match	2024-04-17 22:21:41 +01:00
Alex Young	97a4ebdf15	Construct an explicit logger rather than using the root logger	2024-04-17 21:58:24 +01:00
Alex Young	2ba48bcbf4	Merge branch 'main' into ayo-logging-fixes	2024-04-17 20:47:55 +01:00
Dominik Macháček	bb93952fd2	Merge branch 'main' into online-from-factory	2024-04-17 15:07:00 +02:00
Dominik Macháček	e0f5d42b13	better documentation, help message and logging prints	2024-04-17 14:51:49 +02:00
Alex Young	2afc97db48	Set the log level inside faster-whisper again (lost in merge)	2024-04-14 20:16:28 +01:00
Alex Young	a7cb7a5469	Merge branch 'main' into ayo-logging-fixes	2024-04-14 20:09:56 +01:00
Alex Young	380c30d48d	Further tidying of print output, so by default there's little on the console	2024-04-14 19:14:56 +01:00
Tijs Zwinkels	bccbb15177	Move creation of OnlineASRProcessor inside the factory method ... Preventing more code duplication between whisper_online.py and whisper_online_server.py	2024-03-20 16:29:01 +01:00
Tijs Zwinkels	006de3e7b0	Fix imports ... Now, the ASR implementations do their own imports. No need to import in the factory	2024-03-20 16:02:24 +01:00
Tijs Zwinkels	8896389ea3	Fix crash when using openai-api with whisper_online_server ... + refactored creation of the ASR into a factory method	2024-03-20 15:29:10 +01:00
koiking213	4405c451ce	specify dtype for librosa.load, instead of cast	2024-02-20 23:29:25 +09:00
koiking213	24926c98e0	specify audio dtype	2024-02-20 22:46:04 +09:00
Dominik Macháček	db8b7d2883	removed unused variable	2024-02-20 14:37:18 +01:00
Aleksei Scripnic	80eb0baf5d	Removed duplicate variable self.last_chunked_at ... I tried to find the difference between self.last_chunked_at and self.buffer_time_offset, and it took me a while to understand that they are exactly the same. I think it's better to get rid of one of the duplicates to make the code more readable.	2024-02-20 14:37:18 +01:00
Dominik Macháček	949304ab05	Merge branch 'opeanai-api2' into opeanai-api	2024-02-19 13:51:26 +01:00
Tijs Zwinkels	9fcd403439	Use automatic language detection by default (instead of English)	2024-02-15 22:24:43 +01:00
Tijs Zwinkels	922ad18ebc	Make OpenAI backend work with language autodetect	2024-02-14 17:29:45 +01:00
Tijs Zwinkels	f0a24cd5e1	Make --vad work with --backend openai-api	2024-02-14 17:01:29 +01:00
Tijs Zwinkels	3696fef2b1	Use OpenAI api word-level timestamps	2024-02-14 17:01:29 +01:00
Tijs Zwinkels	531418ad07	Interpolate word timestamps based on word character length	2024-02-14 17:01:29 +01:00
Dominik Macháček	2270014219	fixes	2024-02-14 17:01:29 +01:00
Dominik Macháček	f8b2ae07b8	missing features in openai-api, PR #52	2024-02-14 17:01:29 +01:00
Tijs Zwinkels	f412812082	OpenAI Whisper API backend	2024-02-14 17:01:29 +01:00
Dominik Macháček	c8123344c6	increasing timestamps fixed ... but the code needs to be simplified and cleaned before merging	2024-02-06 16:38:57 +01:00