github FluidInference/FluidAudio v0.14.2

latest releases: v0.17.4, v0.17.3, v0.17.2...
5 months ago

What's Changed

  • CosyVoice3 Mandarin zero-shot TTS port: experimental backend gated to macOS 15 / iOS 18 (CoreML MLState). Package baseline stays at macOS 14 / iOS 17, so ASR / Diarization / VAD / Kokoro / PocketTTS consumers are unaffected (#536)
  • multi-language support EN + 5 new languages: Each pack also ships an optional selective-int8 flowlm_stepv2.mlmodelc for opt-in size reduction: 6L 145 MB → 74 MB, 24L 1.1 GB → 291 MB. Swift continues to load fp16 by default; clients can swap in the int8 variant (#549)
  • New Kokoro ANE model 7-stage CoreML chain support (#547)
  • Avoid misleading confidence warning in SlidingWindowAsrManager.finish() (#548)

Full Changelog: v0.14.1...v0.14.2

New Contributors

Full Changelog: v0.14.1...v0.14.2

Don't miss a new FluidAudio release

NewReleases is sending notifications on new releases.