v0.5.2

What's Changed

• fix(omnivoice): stop swallowing audio tokenizer load failures by @AmirF194 in #912
• Voxtral: stop on (11), not on the text token 32000 by @happyarts in #901
• Add VibeVoice-ASR-Streaming model support by @lucasnewman in #940
• Fix unknown STT kwargs causing generation failures by @lucasnewman in #939
• docs: add VoxCPM2 to supported models by @Blaizzy in #942
• Add NVIDIA NemotronLabs VoiceChat support by @Lazarus-931 in #886
• Breeze tts segment support by @SharkyRawr in #941
• docs: add VoxCPM2 model guide by @Blaizzy in #943
• Add DialogueSidon speaker separation model by @lucasnewman in #948
• Add MiMo-Audio STS model & codec by @lucasnewman in #947
• Bump version to 0.5.2 by @lucasnewman in #949

New Contributors

• @AmirF194 made their first contribution in #912
• @happyarts made their first contribution in #901

Full Changelog: v0.5.1...v0.5.2