diffusion文件夹内为whisper_ppg encoder下训练到420ksteps的预训练模型
另外两个文件夹为whisper_ppg encoder下的sovits预训练模型，300ksteps为目标步数，264ksteps为loss低点，配置文件中的‘n_speakers’应修改为132
使用数据集vctk+m4singers+opencpop+kiritan
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support
HF Inference deployability: The model has no library tag.