Amphion
Open-source toolkit for reproducible audio, music and speech generation (TTS, SVS, VC, TTA).
| Language | Python |
|---|---|
| Category | ML & generative |
| License | MIT |
| Platforms | Linux |
| Install | git clone the repo and build the conda env via env.sh |
| First released | 2023 |
| Maintained | Yes |
| Links | Source |
Strengths
- Broad coverage: TTS, singing synthesis, voice conversion, vocoders
- Reproducible research recipes and pretrained models
- Actively developed
Limitations
- Heavy setup; GPU/CUDA effectively required
- Recipe-driven research workflow, not plug-and-play
Best for
Research and experimentation across speech, singing and audio generation models.