search

THUDM / GLM-4-Voice

GLM-4-Voice | 端到端中英语音对话模型
Apache License 2.0
1.96k stars 144 forks source link

Choose Speaker voice #21

Closed matbee-eth closed 3 days ago

matbee-eth commented 6 days ago

Can we choose what voice the output uses?

darkacorn commented 6 days ago

not via code - probably needs finetuneing

arch-btw commented 6 days ago

This reply might help: https://github.com/THUDM/GLM-4-Voice/issues/1#issuecomment-2437444666

It translates to:

"You can change the tone by fine-tuning the Flow Matching model. At this stage, you can refer to the training code in CosyVoice to train by yourself. In the future, you may consider opening up the Flow Matching fine-tuning method."