Spaces:
Running
Running
Commit History
9 days ago
-
fix: keep API parameter names, reject unknown voices, and stop cloned Chatterbox voices reaching other users Running
Michael Hu4d9637e -
refactor: move model loading and WAV output into shared TTS and STT base classes
Michael Hucf1c514 -
feat: add Cactus Whistle speech recognition tab with telemetry turned off
Michael Huc829773 -
refactor: delete the unused Dia TTS source files
Michael Hubde42b3 -
refactor: remove the dead Dia TTS tab and its factory entry
Michael Hu7de9251 -
fix: guard KittenTTS 2 reference transcription and pin gradio 6.8.0 explicitly
Michael Hu666bdaf -
feat: add KittenTTS 2 tab with built-in voices, voice cloning and expression presets
Michael Hud8fa107 -
fix: repair Piper, KittenTTS, Kokoro and Whisper wrappers and show errors in the UI
Michael Hu19f22b1
-
fix: update Kokoro TTS model to use new pipeline interface and 24 kHz sample rate
Michael Hu3ab544b -
fix: replace Kokoro model's TTS call with direct waveform generation
Michael Hu97fe901 -
fix: ensure Kokoro TTS model is initialized before use and handle missing audio file edge case
Michael Hu9b4a9ad -
refactor: replace inline model definitions with ModelFactory and remove unused imports
Michael Hu- Remove all hard-coded model definitions and import the new factory - Delete unused imports (torchaudio, sys, soundfile, transformers, etc.) - Eliminate duplicate code for model discovery and voice handling - Delete unused utility functions and duplicate code paths - Remove unused dependency on librosa and soundfileef4db28