Commit graph

  • 3192553e20 fixes #307 Quentin Fuxa 2025-12-08 23:52:00 +0100
  • f379a243fe
    Merge pull request #274 from blakkd/patch-1 Quentin Fuxa 2025-12-09 10:10:32 +0100
  • ec09898a9f fixes #301 Quentin Fuxa 2025-12-06 10:19:50 +0100
  • befbae56c7
    minor path change blakkd 2025-11-16 23:47:58 +0100
  • 719e8b1a20 adapt online for mlx detection Quentin Fuxa 2025-11-25 23:52:00 +0100
  • f1b47178d8 adapt online for mlx detection Quentin Fuxa 2025-11-25 23:52:00 +0100
  • 59db08e961 loader for full mlx Quentin Fuxa 2025-11-25 23:52:00 +0100
  • 6fc20b9562 new dec class Quentin Fuxa 2025-11-21 23:52:00 +0100
  • fac8659161 uses native mlx function for attention Quentin Fuxa 2025-11-21 23:52:00 +0100
  • 4d9332ce7d fixes #299 Quentin Fuxa 2025-12-05 17:54:14 +0100
  • 62444ce746 session parameter required in OnnxWrapper Quentin Fuxa 2025-12-05 15:37:18 +0100
  • 2431a6bf91 isolated VAD states per user: .onnx: share a stateless model. .jit: require duplicating the model. Quentin Fuxa 2025-12-05 15:27:14 +0100
  • d1263e7228
    Merge pull request #308 from gzz2000/main Quentin Fuxa 2025-12-05 11:34:05 +0100
  • 30ddd522a4
    Fix local agreement backend, removing excess parameter, fixes https://github.com/QuentinFuxa/WhisperLiveKit/issues/295 Zizheng Guo 2025-12-04 16:45:23 +0800
  • 635bace09e update archi Quentin Fuxa 2025-11-30 18:39:10 +0100
  • f1113e3eb0 update with LoRA Quentin Fuxa 2025-11-29 18:33:30 +0100
  • cc5f819ce7 hf weights Quentin Fuxa 2025-11-29 17:50:46 +0100
  • 82cd24bb75 LoRa path v0 - functional Quentin Fuxa 2025-11-29 17:21:10 +0100
  • d45c397c6a simulstreaming: limit n tokens to prevent hallucinations Quentin Fuxa 2025-11-28 21:41:19 +0100
  • 45bf3f57d7 troubleshooting doc for aarch64 systems Quentin Fuxa 2025-11-28 21:40:43 +0100
  • 1d88ba9d69 Fixes #294. improve model path backend detection and file extraction Quentin Fuxa 2025-11-27 23:14:00 +0100
  • c0965c6c31 Lines to Segments. Merging dataclasses Quentin Fuxa 2025-11-27 21:54:58 +0100
  • 34ddd2ac02 update doc Quentin Fuxa 2025-11-25 23:20:00 +0100
  • 345d781e97 update doc Quentin Fuxa 2025-11-25 23:20:00 +0100
  • 28cf831701 indicate for context token limits for --max-context-tokens. bump to 0.2.16.dev0 Quentin Fuxa 2025-11-25 23:45:15 +0100
  • 60c62f8f84 troubleshooting #271 #276 #284 #286 Quentin Fuxa 2025-11-25 23:31:46 +0100
  • 7faa21f95f alignatt: enable model sharing by removing hooks and centralizing session state. Solves #282 Co-authored-by: Emmanuel Schmidbauer <eschmidbauer@gmail.com> Quentin Fuxa 2025-11-25 23:07:42 +0100
  • 4e9f951551 correct silences handling when language not auto Quentin Fuxa 2025-10-23 11:20:00 +0200
  • 870141298c isort Quentin Fuxa 2025-11-23 11:20:00 +0100
  • 872faa422a correct silences handling when language not auto Quentin Fuxa 2025-10-23 11:20:00 +0200
  • fc9cb66813 disabling vac is not advised Quentin Fuxa 2025-11-23 11:20:00 +0100
  • a175d1a327 fixes silence detected but never reported by silero Quentin Fuxa 2025-11-23 11:20:00 +0100
  • 6206fff118 0.2.15 Quentin Fuxa 2025-11-21 23:52:00 +0100
  • b5067249c0 stt/diar/nllw alignment: internal rework 5 Quentin Fuxa 2025-11-20 23:52:00 +0100
  • f4f9831d39 stt/diar/nllw alignment: internal rework 5 Quentin Fuxa 2025-11-20 23:52:00 +0100
  • 254faaf64c stt/diar/nllw alignment: internal rework 5 Quentin Fuxa 2025-11-20 23:52:00 +0100
  • 8e7aea4fcf internal rework 4 Quentin Fuxa 2025-11-20 23:45:20 +0100
  • 270faf2069 internal rework 3 Quentin Fuxa 2025-11-20 22:28:30 +0100
  • b7c1cc77cc internal rework 2 Quentin Fuxa 2025-11-20 22:06:38 +0100
  • 9a45ec221c internal rework 1 Quentin Fuxa 2025-11-20 12:58:38 +0100
  • 3e13ee6fc3 bump to post4 Quentin Fuxa 2025-11-19 21:23:43 +0100
  • b7d20a0ff0 segment attribution in result formatter Quentin Fuxa 2025-11-19 21:10:28 +0100
  • c1bb9c2bde reduce flickering remaining_time_transcription Quentin Fuxa 2025-11-19 19:09:37 +0100
  • 11e9def0b2 diarization corrections Quentin Fuxa 2025-11-19 19:06:03 +0100
  • 3104f40f6e fixes #279 #278 Quentin Fuxa 2025-11-19 18:17:50 +0100
  • e9b4ceeee5 Add audio partial silence in chunks handling. bump to 0.2.14.post3 Quentin Fuxa 2025-11-17 22:52:00 +0100
  • 437641fb43 reduce min-chunk-size to 0.1, set default model to base Quentin Fuxa 2025-04-27 23:52:00 +0200
  • bfd60b3921 Add audio partial silence in chunks handling. bump to 0.2.14.post2 Quentin Fuxa 2025-11-17 22:52:00 +0100
  • 1e67bf97f0 improve buffering when use of heavy models Quentin Fuxa 2025-04-27 23:52:00 +0200
  • bbd4fd6cff Merge branch 'improve_EOS_handling' Quentin Fuxa 2025-11-16 22:30:31 +0100
  • 28985962a0 Silence handling: finish transcription even if not validated at the BEGINNING of the silence Quentin Fuxa 2025-11-16 22:29:08 +0100
  • a38c103fcd simulstreaming coreml encoder compatibility Quentin Fuxa 2025-11-16 21:24:14 +0100
  • 4d2ffb24f8 coreml conversion Quentin Fuxa 2025-11-16 19:11:43 +0100
  • 1bbbb7903c lora loader in shared whisper core Quentin Fuxa 2025-11-16 18:44:35 +0100
  • bcffdbc6b3 bump to 0.2.14 Quentin Fuxa 2025-11-15 20:19:09 +0100
  • 80b77998f9 Refactor backend handling Quentin Fuxa 2025-11-15 19:51:41 +0100
  • d310f7e25f hf compatibility Quentin Fuxa 2025-11-15 18:34:19 +0100
  • 8d9be88fe6 translation buffer is now displayed in frontend Quentin Fuxa 2025-11-10 15:22:26 +0100
  • 16461052ed task to direct-english-translation Quentin Fuxa 2025-11-10 13:20:26 +0100
  • 5491dbd824 last_validated_token handled in state Quentin Fuxa 2025-11-10 13:18:52 +0100
  • 13401ffe24 whisper core at root of wlk Quentin Fuxa 2025-11-10 12:17:18 +0100
  • 7108d2ddc5 fixes https://github.com/QuentinFuxa/WhisperLiveKit/issues/269 Quentin Fuxa 2025-11-09 20:08:18 +0100
  • a732e0903e Add a script to detect alignement heads, usefull for distilled whisper Quentin Fuxa 2025-11-09 18:12:09 +0100
  • 0491681be4 Distilled model compatibility with HF config.json to ModelDimensions Quentin Fuxa 2025-11-08 20:20:05 +0100
  • ffe5284764 _processing_tasks_done checks task completion Quentin Fuxa 2025-11-05 23:34:00 +0100
  • 41ca17acda to 0.2.13 Quentin Fuxa 2025-10-30 23:30:49 +0100
  • 06b31f51eb exception when translation and no nllw Quentin Fuxa 2025-10-30 23:30:19 +0100
  • ece02db6a3 Use optional new separate NLLW package for translation Quentin Fuxa 2025-10-30 19:36:28 +0100
  • 939a7ebf8b Translation Local Agreement + Cache optimization v0. Not connected yet Quentin Fuxa 2025-10-28 00:16:52 +0100
  • 61edb70fff audioProcessor state variables are now uniquely in State dataclass Quentin Fuxa 2025-10-26 18:54:47 +0100
  • 4e455b8aab translation now separates validated from output buffer tokens Quentin Fuxa 2025-10-26 18:51:09 +0100
  • 9434390ad3 simplify task stopping condition Quentin Fuxa 2025-10-26 17:26:43 +0100
  • 65250db92c tensor to list at the stream end Quentin Fuxa 2025-10-26 16:40:12 +0100
  • 416dce7975 fixes #261 Co-authored-by: yosagi <11404771+yosagi@users.noreply.github.com>" Quentin Fuxa 2025-10-25 14:20:08 +0200
  • 0c5365e7c6 fixes #258 Quentin Fuxa 2025-10-24 20:51:16 +0200
  • 19e9d76610 fixes #257 Quentin Fuxa 2025-10-24 20:39:37 +0200
  • e7b05b0138 migration to silero vad v6: supports onnx Quentin Fuxa 2025-10-23 23:52:00 +0200
  • 818c9c37ca README: path to doc for model file format Quentin Fuxa 2025-10-23 20:34:36 +0200
  • 714fb3b14a custom faster-whisper/mlx whisper encoder available Quentin Fuxa 2025-10-23 20:33:17 +0200
  • 0af379c465 DOC: information about file format Quentin Fuxa 2025-10-23 20:32:05 +0200
  • 9c5bb5df19 README: dir to pah Quentin Fuxa 2025-10-23 20:31:12 +0200
  • dc6ea79036 apache license inheritance from simulwhisper and nemo Quentin Fuxa 2025-10-23 20:28:02 +0200
  • 21bbb59e31
    Merge pull request #250 from ladinu/patch-1 Quentin Fuxa 2025-10-15 08:59:02 +0200
  • 12a69205ed bump to 0.2.12 Quentin Fuxa 2025-10-06 19:59:05 +0200
  • 1f684cdd97 fixes #251 Quentin Fuxa 2025-10-06 19:53:27 +0200
  • 3467109668
    fix broken link Ladinu Chandrasinghe 2025-10-05 10:51:41 -0700
  • 290470dd60 forwarded_allow_ips in core Quentin Fuxa 2025-10-04 23:04:00 +0200
  • 425ac7b51d forwarded_allow_ips in core Quentin Fuxa 2025-10-04 23:04:00 +0200
  • 0382cfbeba forwarded_allow_ips in core Quentin Fuxa 2025-10-04 23:04:00 +0200
  • 9b1e061b32 forwarded_allow_ips in core Quentin Fuxa 2025-10-04 23:04:00 +0200
  • b4abc158b9
    Merge pull request #249 from Damrod/add-ip-forwarding-support Quentin Fuxa 2025-10-06 10:20:05 +0200
  • 5832d7433d update documentation Alvaro Ollero 2025-10-04 23:18:10 +0200
  • 3736458503 Uvicorn exposes a configuration option to enable reverse proxying from a trusted ip. This PR exposes it downstreams to end clients Alvaro Ollero 2025-10-04 22:21:06 +0200
  • 374618e050 token speakers are only reattributed for token coming after last_validated_token Quentin Fuxa 2025-10-04 09:52:00 +0200
  • 543972ef38 fixes #248 Quentin Fuxa 2025-10-04 09:52:00 +0200
  • 971f8473eb update api doc Quentin Fuxa 2025-10-05 11:09:47 +0200
  • 8434ef5efc update api Quentin Fuxa 2025-10-05 11:09:12 +0200
  • 73f36cc0ef v0 doc new api Quentin Fuxa 2025-10-02 23:04:00 +0200
  • a7db39d999 solves incorrect spacing in buffer diarization Quentin Fuxa 2025-10-02 23:04:00 +0200
  • a153e11fe0 update when self.diarization_before_transcription Quentin Fuxa 2025-09-28 11:04:00 +0200