File size: 303 Bytes
ef4d549
 
 
 
 
1
2
3
4
5
Pre-trained Bert-VITS2-2.2-CLAP model.

This model is trained with Chinese / Japanese / English speech data, the speaker embedding is removed due to copyright issues.

You can use auxiliary data to fine-tune the model to avoid catastrophic forgetting of multi-language ability and CLAP control ability.