Download Latest Version
v0.4.7 source code.zip (7.2 MB)
Get an email when there's a new version of MLX-Audio
| Name | Modified | Size | Downloads / Week |
|---|---|---|---|
| Parent folder | |||
| README.md | 2026-06-06 | 7.8 kB | |
| v0.4.4 source code.tar.gz | 2026-06-06 | 6.6 MB | |
| v0.4.4 source code.zip | 2026-06-06 | 7.0 MB | |
| Totals: 3 Items | 13.7 MB | 0 | |
What's Changed
- Implement an ICL cache for Qwen3-TTS. by @orbitalquark in https://github.com/Blaizzy/mlx-audio/pull/685
- feat(higgs_audio): overlap-add mid-generation streaming by @Kairos-a in https://github.com/Blaizzy/mlx-audio/pull/669
- Add Step-Audio 2 codec by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/686
- Allow reference voice matching in Qwen3 TTS batch generation by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/689
- Delete XCODE_BUILD_TROUBLESHOOTING.md by @Blaizzy in https://github.com/Blaizzy/mlx-audio/pull/692
- feat: Add VoxCPM2 TTS model (2B params, 48kHz, 30 languages) by @acul3 in https://github.com/Blaizzy/mlx-audio/pull/641
- Qwen3-TTS: honor caller-provided max_tokens in instruct and ICL paths. by @contrapuntal in https://github.com/Blaizzy/mlx-audio/pull/695
- Add MOSS TTS models (delay and local transformer) by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/691
- Fix regression in Fish S2 Pro by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/693
- Add Silero VAD model by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/701
fix(voxtral_realtime): empty downsample_and_project uses decoder dimby @bsmith925 in https://github.com/Blaizzy/mlx-audio/pull/700- feat(stt): OpenAI-compatible response_format on /v1/audio/transcriptions by @mbailey in https://github.com/Blaizzy/mlx-audio/pull/704
- fix: quantize Chatterbox ve projection for 4-bit checkpoints by @masterbatcoderman10 in https://github.com/Blaizzy/mlx-audio/pull/707
- Add MOSS-TTSD dialogue model by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/698
- Fix slow and flaky tests by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/705
- STS voice pipeline updates by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/708
- Cohere ASR: 1.7× faster long-form, multi-batch correctness, VAD pre-processing by @beshkenadze in https://github.com/Blaizzy/mlx-audio/pull/697
- fix(granite-speech): transpose pointwise conv weights from PyTorch layout by @nneubacher in https://github.com/Blaizzy/mlx-audio/pull/715
- Add style instruction support for Fish Speech S2 Pro by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/713
- Fix mx.random.seed() no-op in Fish and Whisper samplers by @emmilco in https://github.com/Blaizzy/mlx-audio/pull/721
- test(irodori-tts): add v3 tests (duration predictor, Sway Sampling) by @yoshphys in https://github.com/Blaizzy/mlx-audio/pull/728
- fix: prevent AudioPlayer hang when short text falls below min_buffer_seconds threshold by @jeon30c in https://github.com/Blaizzy/mlx-audio/pull/727
- feat(vad): Add FSMN-VAD model support by @tian-sweetaylor in https://github.com/Blaizzy/mlx-audio/pull/729
- Pre-flight model load on streaming audio routes by @guygrigsby in https://github.com/Blaizzy/mlx-audio/pull/725
- Fix handling of temperature / min-p / top-p when sampling from Qwen3-TTS by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/735
- Fix error handling for models with periods in the repo ID by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/734
- Add Dramabox model by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/722
- Add KittenTTS to supported model docs by @dewana-sl in https://github.com/Blaizzy/mlx-audio/pull/741
- Fix Kokoro usage from worker threads by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/745
- Add Mega-ASR STT model (Qwen3-ASR-1.7B + audio-quality router + LoRA switching) by @beshkenadze in https://github.com/Blaizzy/mlx-audio/pull/740
- Add server-side VAD turn detection to /v1/realtime by @maltyxx in https://github.com/Blaizzy/mlx-audio/pull/748
- Add support for MOSS-TTS 1.5 by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/749
- feat(stt): add granite-speech-4.1-2b-nar (non-autoregressive ASR) by @mouddane in https://github.com/Blaizzy/mlx-audio/pull/738
- Fix VoxCPM2 Chinese tokenization by @qiuhq-9527 in https://github.com/Blaizzy/mlx-audio/pull/751
- Update README.md by @Saucken1945 in https://github.com/Blaizzy/mlx-audio/pull/752
- Use a dynamic batch size for Cohere Transcribe by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/711
- feat(stt): expose word_timestamps form field on /v1/audio/transcriptions by @ciekawy in https://github.com/Blaizzy/mlx-audio/pull/716
- Add OmniVoice to README by @Blaizzy in https://github.com/Blaizzy/mlx-audio/pull/754
- Add GET /v1/audio/voices endpoint for TTS voice discovery by @etherious1804 in https://github.com/Blaizzy/mlx-audio/pull/743
- Add support for Miso TTS by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/764
- feat(irodori-tts): add Irodori-TTS v3 VoiceDesign dual conditioning (speaker + caption) by @yoshphys in https://github.com/Blaizzy/mlx-audio/pull/759
- Fix LFM2.5-Audio EOS handling by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/768
- feat(stt): add NVIDIA Nemotron 3.5 ASR (streaming) support by @ARahim3 in https://github.com/Blaizzy/mlx-audio/pull/771
- Add Higgs Audio v3 TTS model by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/770
- Fix Canary-1B-v2 loading for MLX-native checkpoints by @evanqhuang in https://github.com/Blaizzy/mlx-audio/pull/763
- nemotron_asr: cache-aware streaming (stream_generate) by @beshkenadze in https://github.com/Blaizzy/mlx-audio/pull/774
- refactor(stt): add shared nemo/ package; decouple nemotron_asr from parakeet by @beshkenadze in https://github.com/Blaizzy/mlx-audio/pull/775
- fix(utils): sharpen resampler anti-aliasing filter by @beshkenadze in https://github.com/Blaizzy/mlx-audio/pull/776
- Update version v0.4.4 by @Blaizzy in https://github.com/Blaizzy/mlx-audio/pull/778
- Add an STT eval harness by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/777
- Fix Whisper best-of-N decoding by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/767
- Add Fun-ASR-Nano model support by @lucasnewman in https://github.com/Blaizzy/mlx-audio/pull/760
New Contributors
- @acul3 made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/641
- @bsmith925 made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/700
- @mbailey made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/704
- @masterbatcoderman10 made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/707
- @nneubacher made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/715
- @emmilco made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/721
- @jeon30c made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/727
- @tian-sweetaylor made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/729
- @guygrigsby made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/725
- @dewana-sl made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/741
- @maltyxx made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/748
- @mouddane made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/738
- @qiuhq-9527 made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/751
- @Saucken1945 made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/752
- @ciekawy made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/716
- @etherious1804 made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/743
- @ARahim3 made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/771
- @evanqhuang made their first contribution in https://github.com/Blaizzy/mlx-audio/pull/763
Full Changelog: https://github.com/Blaizzy/mlx-audio/compare/v0.4.3...v0.4.4