Fetching metadata from the HF Docker repository...
-
Port pocket-tts to supertonic3: 44100 Hz, 10 voices, 31 languages
gavanduffy
Complete rewrite of pocket-tts-openai-server to use supertonic3 TTS engine.
Key changes:
- Replace PyTorch/torchao stack with ONNX Runtime via supertonic3
- 44100 Hz sample rate (was 24000 Hz)
- 10 built-in voices M1-M5/F1-F5 (no voice cloning)
- 31 languages, single model (no per-language switching)
- Default port 7860 (was 49112)
- Simplified streaming via sentence splitting
- Removed quantization, safetensors, voice cache
-
restructure and dockerize
