All pull requests

Pull requests

• fix(profiles): render a design voice's sample with its saved seed

  #3382·debpalash opened Oct 25, 2026Owner·

  ·
  1

  4

• fix(sherpa-onnx): load Piper voices and lexicon VITS models

  #3373·MehmetinLove opened Oct 25, 2026·14 tasks·

  ·
  12

• fix(captions): omit ruby annotations from spoken imports

  #3360·014-code opened Oct 25, 2026·6 of 7 tasks·

  ·
  1

  8

• fix(tts): speak the trailing zeros written in a percentage

  #3359·kevin9327 opened Oct 25, 2026Contributor·2 tasks done·

  ·
  3

• fix(clone): preserve manual reference text for saved voices

  #3358·isnaenihidayat opened Oct 25, 2026·7 of 8 tasks·

  ·
  1

  6

• feat(backend-gate): one progress card with a step checklist for the first-run install

  #3350·meisolated opened Oct 25, 2026·5 of 14 tasks·

  ·
  12

• fix(perf): use MLX Whisper for Apple Silicon model packs

  #3349·bee-interactive opened Oct 25, 2026·5 of 6 tasks·

  ·
  6

• feat(clone): add IndexTTS2 emotion controls to Voice Cloning

  #3344·XiaoPuOuO opened Oct 25, 2026·

  ·
  8

• fix(design): preserve free-form prompts in saved voice profiles

  #3343·XiaoPuOuO opened Oct 25, 2026·

  ·
  19

• fix(profiles): keep a locked take's whole text as its reference transcript

  #3320·bercedev opened Oct 24, 2026Contributor·3 of 7 tasks·

  ·
  2

• fix: consolidate 98 community fixes from @rudycelekli (replaces 98 small PRs)

  #3262·velixio opened Oct 24, 2026Collaborator·

  ·
  11

• Add native CosyVoice GGUF Vulkan engine

  #3249·dutch2005 opened Oct 24, 2026·

  ·
  1

  11

• fix(tts): strip markdown emphasis before synthesis (#3197)

  #3205·swadhinbiswas opened Oct 23, 2026Contributor·

  ·
  1

  25

• fix(dub): keep Mod+Z and Mod+S from editing segments outside segment fields

  #3199·HEOJUNFO opened Oct 23, 2026·5 of 13 tasks·

  ·
  2

• fix(backend): guard torchaudio import and register PyTorch DLL paths (#3147)

  #3186·JaiPrakashVI opened Oct 22, 2026·7 of 13 tasks·

  ·
  1

  15

• test(backend): pin transient health-check recovery from issue #3184

  #3185·nirmalchatur opened Oct 22, 2026Contributor·

  ·
  1

  4

• fix(asr): honor selected engine for reference transcription (#3176)

  #3183·nirmalchatur opened Oct 22, 2026Contributor·

  ·
  1

  21

• fix(moss-tts-nano): preserve configured tokenizer caches

  #3167·ZFzuofeng opened Oct 22, 2026·5 tasks done·

  ·
  1

  8

• fix(tests): resolve bun before spawning it so Windows collection cannot abort

  #3155·shubhankarkanase90-byte opened Oct 22, 2026·7 of 13 tasks·

  ·
  7

• fix(export): stop karaoke sweeps drifting on long captions

  #3134·kevin9327 opened Oct 21, 2026Contributor·4 tasks done·

  ·
  7

• fix(colab): handle OmniVoice licence errors clearly (#3076)

  #3133·nirmalchatur opened Oct 21, 2026Contributor·

  ·
  1

  4

• fix(models): skip native OmniVoice preload when another engine is selected

  #3127·isnaenihidayat opened Oct 21, 2026·6 of 7 tasks·

  ·
  1

  5

• docs: add Korean README (README_KO.md)

  #3118·writer0713 opened Oct 21, 2026·5 tasks done·

  ·
  3

• fix(gpu): name a card the installed PyTorch build has no kernels for

  #3104·Faisal-Fayaz opened Oct 20, 2026·6 of 13 tasks·

  ·
  23

• fix(gguf): ship a runtime that runs on every x86-64 CPU (#2780)

  #3075·inteligencia66 opened Oct 20, 2026·5 of 13 tasks·

  ·
  1

  8