Zamia-TTS is an experimental project that tries to train TTS voices based on (reviewed) Zamia-Speech data.
Currently, two TTS engines are supported:
- Tacotron (lower quality but also lower hardware requirements), code here: https://github.com/gooofy/zamia-tts
- Tacotron-2 (higher quality but will require significant GPU power), code here: https://github.com/NVIDIA/tacotron2
Voice training recipies can be found inside the Zamia-Speech README here: