Skip to content

v0.7.3

Latest

Choose a tag to compare

@github-actions github-actions released this 07 Sep 22:46
· 20 commits to main since this release
9c6a282

What's Changed

  • ggml: additive INT8/ternary types, fused CPU ops, and kernels for VibeASR (1/2) by @XsquirrelC in #447
  • server: add /v1/audio/transcriptions/details for transcript detail fields by @CryptVenture in #401
  • [Feature] ModelScope download by @IIIIIllllIIIIIlllll in #450
  • sanotts: sanoTTS voice family community model (seven voices, 294k-2.27M params, en/vi/id, GGUF FP32) by @voidash in #449
  • higgs_audio_tts: give the codec decoder right-hand context at the end of the stream by @niksedk in #454
  • Add PocketTTS streaming support by @0xShug0 in #458
  • community_models: VibeASR INT8/ternary ASR pipeline (2/2) by @XsquirrelC in #448
  • Add 'Ctrl' variant to breeze_tts entry by @Th-Underscore in #463
  • gguf: let a finished GGUF be re-converted by @CaptainArni in #464
  • Add Vulkan Docker image support by @0xShug0 in #466
  • Add CLI and server version output by @0xShug0 in #469
  • Inject build version in CI and release packages by @0xShug0 in #470
  • feat: add native MiraTTS community model by @mirek190 in #395
  • feat(model): add Sopro V2 Turbo (sopro_tts) with offline and streaming by @Brioch in #353
  • fix(attention): fallback for GPUs without flash MMA kernels (sm70) by @Th-Underscore in #423
  • Fix Chatterbox multilingual diacritic normalization by @0xShug0 in #468
  • Use ViewKV flash attention for Qwen prefix prefill by @0xShug0 in #477
  • Add VoxCPM2 stateless chunk strategy by @0xShug0 in #478
  • Buffer chunked HTTP uploads by @0xShug0 in #479
  • fix(build): guard CUDA driver probe out of HIP builds by @IIIIIllllIIIIIlllll in #480
  • Add Vevo2 source audio chunking by @0xShug0 in #482
  • ci: build and run the unit test suite on Linux/macOS/Windows by @gqf2008 in #442
  • Tighten BreezeTTS streaming path by @0xShug0 in #483

New Contributors

Full Changelog: v0.7.2...v0.7.3