- Turned the acoustic synthesizer's notebook into scripts in the format of classic STTS repos
- Saving, Logging and Loading the checkpoints so far seem to be working good
- added the config to the training script
- Only DP so far, may add Accelerate later
- pyannote's SV didn't work, replaced it with another instance of WavLM SV with a modified SVLoss.
- Python >= 3.7
- Clone this repository:
git clone https://github.com/SoshyHayami/StyleTTS-ZS.git
cd StyleTTS2
- Install python requirements:
pip install -r requirements.txt
acousting synth training:
python train_.py --config_path ./Configs/config.yml