Parallel WaveGAN fork for Mozilla TTS
This is a fork of the original implementation to sync with Mozilla TTS.
| Pre-Trained Model | Dataset | Commit | Details |
|---|---|---|---|
| PWGAN | LJSpeech | fca88f9 | Trained with GT spectrograms |
| MelGAN | LJSpeech | fe9c02e | Trained with GT spectrograms |
Basic workflow
- Create spectrograms using TTS based audio processing python python bin/preprocess_with_ap.py (this requires Mozilla TTS to be installed).
- Setup the config file wrt audio processing parameters used in your TTS model.
- Train the model.
Visit the original Repo for more information.