admin-valentin/WhisperLiveKit
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Commit graph

  • 9a8d3cbd90 improve diarization + silence handling Quentin Fuxa 2024-08-24 19:20:00 +0200
  • b101ce06bd several users share the same sortformer model instance Quentin Fuxa 2024-08-24 19:18:00 +0200
  • c83fd179a8 improves phase shift correction between transcription and diarization Quentin Fuxa 2024-08-24 19:15:00 +0200
  • 5258305745 default diarization backend in now sortformer Quentin Fuxa 2025-08-24 18:32:01 +0200
  • ce781831ee punctuation is checked in audio-processor's result formatter Quentin Fuxa 2025-08-24 18:32:01 +0200
  • 58297daf6d sortformer diar implementation v0.3 Quentin Fuxa 2025-08-24 18:32:01 +0200
  • 3393a08f7e sortformer diar implementation v0.2 Quentin Fuxa 2025-08-24 18:32:01 +0200
  • 5b2ddeccdb correct pip installation error in image build Quentin Fuxa 2025-08-22 15:37:46 +0200
  • 26cc1072dd new dockerfile for cpu only. update dockerfile from cuda 12.8 to 12.9 Quentin Fuxa 2025-08-22 11:04:35 +0200
  • 12973711f6 0.2.6 Quentin Fuxa 2025-08-21 14:34:46 +0200
  • 909ac9dd41 speaker -1 are no more sent in websocket - no buffer when their is a silence Quentin Fuxa 2025-08-21 14:09:02 +0200
  • d94a07d417 default model is now base. default backend simulstreaming Quentin Fuxa 2025-08-21 11:55:36 +0200
  • b32dd8bfc4 Align backend and frontend time handling Quentin Fuxa 2025-08-21 10:33:15 +0200
  • 9feb0e597b remove VACOnlineASRProcessor backend possibility Quentin Fuxa 2025-08-20 20:57:43 +0200
  • 9dab84a573 update front Quentin Fuxa 2025-08-20 20:15:38 +0200
  • d089c7fce0 .html to .html + .css + .js Quentin Fuxa 2025-08-20 20:00:31 +0200
  • 253a080df5 diart diarization handles pauses/silences thanks to offset Quentin Fuxa 2025-08-19 21:12:55 +0200
  • 0c6e4b2aee sortformer diar implementation v0.1 Quentin Fuxa 2025-08-19 19:48:51 +0200
  • e14bbde77d sortformer diar implementation v0 Quentin Fuxa 2025-08-19 17:02:55 +0200
  • 7496163467 rename diart backend Quentin Fuxa 2025-08-19 15:02:27 +0200
  • 696a94d1ce 1rst sortformer backend implementation Quentin Fuxa 2025-08-19 15:02:17 +0200
  • 2699b0974c Fix simulstreaming imports Quentin Fuxa 2025-08-19 14:43:54 +0200
  • 90c0250ba4 update optional dependencies Quentin Fuxa 2025-08-19 09:36:59 +0200
  • eb96153ffd
         new vac parameters Quentin Fuxa 2025-08-17 22:26:28 +0200
  • 47e3eb9b5b
         Update README.md Quentin Fuxa 2025-08-17 09:55:03 +0200
  • b8b07adeef --vac to --no-vac Quentin Fuxa 2025-08-17 09:44:26 +0200
  • d0e9e37ef6 simulstreaming: cumulative_time_offset to keep timestamps correct when audio > 30s Quentin Fuxa 2025-08-17 09:33:47 +0200
  • 820f92d8cb audio_max_len to 30 -> 20, ffmpeg timeout 5 -> 20 Quentin Fuxa 2025-08-17 09:32:08 +0200
  • e42523af84 VAC activated by default Quentin Fuxa 2025-08-17 01:29:34 +0200
  • e2184d5e06 better handle silences when VAC + correct offset issue with whisperstreaming backend Quentin Fuxa 2025-08-17 01:27:07 +0200
  • 7fe0353260 vac model is loaded in TranscriptionEngine, and by default Quentin Fuxa 2025-08-17 00:34:25 +0200
  • 0f2eba507e use with_offset to add no audio offset to tokens Quentin Fuxa 2025-08-17 00:33:24 +0200
  • 55e08474f3 recycle backend in simulstreaming thanks to new remove hooks function Quentin Fuxa 2025-08-16 23:06:16 +0200
  • 28bdc52e1d VAC before doing transcription and diarization. V0 Quentin Fuxa 2025-08-16 23:04:21 +0200
  • e4221fa6c3 Merge branch 'main' of https://github.com/QuentinFuxa/whisper_streaming_web Quentin Fuxa 2025-08-15 23:04:05 +0200
  • 1652db9a2d Use distinct backend models for simulstreaming and add --preloaded_model_count to preload them Quentin Fuxa 2025-08-15 23:03:55 +0200
  • 601f17653a
         Update CONTRIBUTING.md Quentin Fuxa 2025-08-13 21:59:32 +0200
  • 7718190fcd
         Update CONTRIBUTING.md Quentin Fuxa 2025-08-13 21:59:00 +0200
  • 349c7dcb9e bump version ro 0.2.5 Quentin Fuxa 2025-08-13 10:04:31 +0200
  • 1c42b867cf Merge branch 'main' of https://github.com/QuentinFuxa/whisper_streaming_web Quentin Fuxa 2025-08-13 10:04:04 +0200
  • d4771e563e Increase END_SILENCE_DURATION to reduce false positives Quentin Fuxa 2025-08-13 10:04:00 +0200
  • b0a5fc0693
         Merge pull request #155 from davidgumberg/keepawakescrolldown Quentin Fuxa 2025-08-13 10:02:52 +0200
  • 3b96fb8776 frontend: Scroll down when appending transcription David Gumberg 2025-08-12 17:31:32 -0700
  • 7f93c4b978 frontend: Don't let screen sleep when transcribing. David Gumberg 2025-08-12 17:30:57 -0700
  • 15c3df1cba warmup base whisper when using simulstreaming Quentin Fuxa 2025-08-12 18:52:52 +0200
  • 7fb8e66c01 typo Quentin Fuxa 2025-08-12 18:36:32 +0200
  • 728e1f1290 simulstreaming warmup is done for each instance of online, not for the backend Quentin Fuxa 2025-08-12 18:35:04 +0200
  • 87b9ed6ecd nonspeech_prob from 1 to 0.5 Quentin Fuxa 2025-08-12 18:34:37 +0200
  • 38b4ebe8ba Handle 3 types of silences: Indicated by whisper, between tokens, and at the end of the input. Display them in the frontend Quentin Fuxa 2025-08-11 17:56:57 +0200
  • d098af3185 each SimulStreamingOnlineProcessor now contains PaddedAlignAttWhisper instance. SimulStreamingASR only contains loaded whisper model Quentin Fuxa 2025-08-11 08:24:14 +0200
  • 4e56130a40 frontend supports dark theme Quentin Fuxa 2025-08-11 08:22:23 +0200
  • 2bbdc70187 lags are now updated every 0.1s Quentin Fuxa 2025-08-09 23:11:05 +0200
  • b678a55f63 remove duplicate file Quentin Fuxa 2025-08-09 23:10:34 +0200
  • 5491964e81 clean SimulStreamingOnlineProcessor initialization + audio processing Quentin Fuxa 2025-08-09 20:16:27 +0200
  • b05297a96d clean simulwhisper backend and online Quentin Fuxa 2025-08-09 18:02:15 +0200
  • 197293e25e refactor(simulstreaming): extract backend + online module into separate files from whisper streaming Quentin Fuxa 2025-08-08 18:07:51 +0200
  • ba41c4ab56 Remove download_simulstreaming_backend Quentin Fuxa 2025-08-08 18:06:40 +0200
  • bda72b8bc0 setup.py to pyproject.toml. Remove <2.0.0 condition on numpy dep Quentin Fuxa 2025-08-03 16:32:31 +0200
  • bb6b9f4cb1 architecture diagram : available backends for whisper streaming & diarization Quentin Fuxa 2025-08-03 12:25:36 +0200
  • e40b5a3ea0 Update architecture diagram Quentin Fuxa 2025-08-02 13:51:15 +0200
  • 4cfed6e98e in MultiHeadAttention and ResidualAttentionBlock include cache_id for compatibility with simulstreaming code Quentin Fuxa 2025-08-02 13:16:58 +0200
  • 687e3dd5e2 update simulstreaming model.py to match the latest version of whisper sources Quentin Fuxa 2025-08-02 13:16:10 +0200
  • e4140cd299 Update Dockerfile to install build-essential and update PyTorch version Quentin Fuxa 2025-08-02 13:08:43 +0200
  • 8e056cbdf2 Upgrade SimulStreaming Whisper core from version 20230918 to 20250625 Quentin Fuxa 2025-08-02 13:06:36 +0200
  • 9dcfb38967
         Update README.md Quentin Fuxa 2025-08-01 18:02:11 +0200
  • 47b9235d70
         Update README.md Quentin Fuxa 2025-08-01 17:55:40 +0200
  • f3cd53a4db
         Update README.md Quentin Fuxa 2025-08-01 16:53:22 +0200
  • dbdb4ea66c
         Update README.md Quentin Fuxa 2025-08-01 16:33:26 +0200
  • 00424d7ca3 latest version of simulstreaming Quentin Fuxa 2025-07-31 16:44:23 +0200
  • 4b738d6f63 fix duplicate line Quentin Fuxa 2025-07-31 16:29:35 +0200
  • 8a5e2adb1e simulstreaming: fixes token handling during warm-up phase Quentin Fuxa 2025-07-31 16:25:34 +0200
  • f85329e112 Merge branch 'main' of https://github.com/QuentinFuxa/whisper_streaming_web Quentin Fuxa 2025-07-31 11:42:16 +0200
  • 46efbdf1d9 solves https://github.com/QuentinFuxa/WhisperLiveKit/issues/151 Quentin Fuxa 2025-07-31 11:42:06 +0200
  • 8885ade003
         Merge pull request #153 from luisla-rivas/main Quentin Fuxa 2025-07-31 07:10:35 +0200
  • 2564928d83 Fix README.md to view correctly Deployment Guide info luisla-rivas 2025-07-30 14:11:19 +0200
  • 56114d3071 Remove end_attributed_speaker in diarization_online. handled in audio processor Quentin Fuxa 2025-07-16 12:09:43 +0200
  • 5b9977c9af Enhanced use_punctuation_split for diarization. further improvements still needed Quentin Fuxa 2025-07-16 12:06:17 +0200
  • 12a544164f Merge branch 'main' of https://github.com/QuentinFuxa/whisper_streaming_web Quentin Fuxa 2025-07-16 12:05:01 +0200
  • 2ca1156b7e
         Merge pull request #147 from choomegan/diar_queue Quentin Fuxa 2025-07-16 12:04:53 +0200
  • 3ad3683ca7 Refactor speaker assignment in DiartDiarization for clarity and punctuation awareness Quentin Fuxa 2025-07-15 14:38:53 +0200
  • 1599bd87a0 work on punctuation_split Quentin Fuxa 2025-07-15 12:04:54 +0200
  • 90623400a4 Remove automatic downloading of SimulStreaming dependencies on import failure Quentin Fuxa 2025-07-15 12:04:17 +0200
  • 64e44fb24f fix: logic of adding of pcm_array to diarization_queue choomegan 2025-07-15 15:27:44 +0800
  • 156b9a133f 0.2.2 Quentin Fuxa 2025-07-04 17:11:35 +0200
  • df8cb23848 Merge branch 'main' of https://github.com/QuentinFuxa/whisper_streaming_web Quentin Fuxa 2025-07-04 17:04:26 +0200
  • 9ff513093b simulstreaming uses empty space as separator Quentin Fuxa 2025-07-04 17:03:01 +0200
  • 17184e552c
         Update README.md Quentin Fuxa 2025-07-03 11:13:45 +0200
  • aad2c55d8c download_simulstreaming_backend.py now downloads files in the correct lib dir Quentin Fuxa 2025-07-03 11:07:28 +0200
  • 2f177c4a3b add __init__.py file to simul_whisper assets directory Quentin Fuxa 2025-07-03 10:41:12 +0200
  • b362eccb23 new command to get simulstreaming backend Quentin Fuxa 2025-07-03 10:24:02 +0200
  • 5daaf77258 add download script for SimulStreaming backend Quentin Fuxa 2025-07-03 10:14:45 +0200
  • 36cc4412c3 update LICENSE with SimulStreaming dual licensing terms; include in .gitignore additional stuff Quentin Fuxa 2025-07-03 09:21:38 +0200
  • e1d4bf7e94 modify import paths in simul whisper backend so that it works in lib mode Quentin Fuxa 2025-07-01 20:34:47 +0200
  • 62bf28949e compatible with the latest version of simulstreaming Quentin Fuxa 2025-07-01 20:10:45 +0200
  • 25526b3aa2 typo Quentin Fuxa 2025-07-01 19:14:49 +0200
  • 1e3fab9550 copy non python files from simulstreaming when installing package Quentin Fuxa 2025-07-01 19:14:23 +0200
  • f25de6d8a4 ffmpeg-python is not used anymore - ffmpeg is directly called through create_subprocess_exec Quentin Fuxa 2025-07-01 18:53:35 +0200
  • 8a175e79d8 Merge branch 'main' of https://github.com/QuentinFuxa/whisper_streaming_web Quentin Fuxa 2025-07-01 18:52:26 +0200
  • dc37b44486 add _read_stderr to empty the stderr Quentin Fuxa 2025-07-01 17:05:58 +0200
  • 2d1df92aa7
         Merge pull request #145 from SlavikCA/port-fix Quentin Fuxa 2025-07-01 14:16:58 +0200