admin-valentin/WhisperLiveKit
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
WhisperLiveKit/whisperlivekit
History
Quentin Fuxa a4da246ea5 feat: add voxtral-mlx native backend for Apple Silicon 
Pure-MLX implementation of Voxtral Mini 4B Realtime for low-latency
speech transcription on Apple Silicon. Avoids the transformers/torch
overhead and runs at 0.18-0.32x real-time factor.

- voxtral_mlx/model.py: MLX model with spectrogram, encoder, decoder
- voxtral_mlx/loader.py: model loading with 6-bit quantized weights
- voxtral_mlx/spectrogram.py: mel spectrogram computation in MLX
- voxtral_mlx_asr.py: VoxtralASR adapter for the AudioProcessor pipeline
2026-02-22 23:28:10 +01:00
..
diarization add insert_audio_chunk to DiartDiarization 2026-02-11 22:10:00 +01:00
local_agreement fix --direct-english-translation not setting task=translate for localagreement backends 2026-02-11 22:10:00 +01:00
silero_vad_models fixes silence detected but never reported by silero 2025-11-23 11:20:00 +01:00
simul_whisper fix: handle numpy object_ dtype from ctranslate2 encoder (#337) 2026-02-20 20:48:28 +01:00
voxtral_mlx feat: add voxtral-mlx native backend for Apple Silicon 2026-02-22 23:28:10 +01:00
web isort 2025-11-23 11:20:00 +01:00
whisper simulstreaming mlx & torch dedup of common base 2025-02-15 23:52:00 +01:00
__init__.py isort 2025-11-23 11:20:00 +01:00
audio_processor.py fix: silence double-counting bug, add metrics module and runtime instrumentation 2026-02-22 23:27:12 +01:00
backend_support.py mixstral hf v0 2026-02-20 20:49:57 +01:00
basic_server.py simulstreaming mlx & torch dedup of common base 2025-02-15 23:52:00 +01:00
config.py simulstreaming mlx & torch dedup of common base 2025-02-15 23:52:00 +01:00
core.py docs: update README with voxtral backend, benchmarks, testing sections 2026-02-22 23:27:57 +01:00
ffmpeg_manager.py isort 2025-11-23 11:20:00 +01:00
metrics.py fix: silence double-counting bug, add metrics module and runtime instrumentation 2026-02-22 23:27:12 +01:00
metrics_collector.py fix: silence double-counting bug, add metrics module and runtime instrumentation 2026-02-22 23:27:12 +01:00
model_mapping.py simulstreaming mlx & torch dedup of common base 2025-02-15 23:52:00 +01:00
model_paths.py Fixes #294. improve model path backend detection and file extraction 2025-11-27 23:14:00 +01:00
parse_args.py docs: update README with voxtral backend, benchmarks, testing sections 2026-02-22 23:27:57 +01:00
silero_vad_iterator.py simulstreaming mlx & torch dedup of common base 2025-02-15 23:52:00 +01:00
thread_safety.py Fix critical thread safety issues 2026-01-09 11:23:19 -05:00
timed_objects.py add probability field to ASRToken 2026-02-11 22:10:00 +01:00
tokens_alignment.py fix NoneType concatenation in add_translation 2026-02-11 22:10:00 +01:00
voxtral_hf_streaming.py correct processor attributes mixtral 2026-02-22 21:13:21 +01:00
voxtral_mlx_asr.py feat: add voxtral-mlx native backend for Apple Silicon 2026-02-22 23:28:10 +01:00
warmup.py isort 2025-11-23 11:20:00 +01:00