Gradient TTS

An implementation of Wang, C., Chen, S., Wu, Y., (2023), Neural Codec Language Models are Zero-Shot Text to Speech Synthesizers

Dataset

Link to download the weights

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

extract common voice to any directory, then fix it lhotse bug

./dataset/fix_common_voice.sh

set args in common_voice.py, then

python -m tts.dataset.common_voice

TODO