2noise
How Can I use my own voice to training the AI
I want to use my own voice to training the AI, but I don't know the steps.Can anyone write a torturial about it?
我不知道怎么使用自己的语音训练AI生成自己的语音，有人能写出详细的步骤吗？
same question, i need the english speaker to speak more singlish/chinese accent
This fork is trying to add fine tuning: https://github.com/ain-soph/ChatTTS
This fork is trying to add fine tuning: https://github.com/ain-soph/ChatTTS
Does it support to train a new language? instead of fine-tune for cloning voice?
This fork is trying to add fine tuning: https://github.com/ain-soph/ChatTTS
Does it support to train a new language? instead of fine-tune for cloning voice?
the fork is not clear how to do the fine tune
This fork is trying to add fine tuning: https://github.com/ain-soph/ChatTTS
Does it support to train a new language? instead of fine-tune for cloning voice?
the fork is not clear how to do the fine tune
first finetune encoder, then finetune gpt
I’m the one who create that fork.
Yes, first train the encoder and then GPT. However, the current encoder loss cannot be optimized to be smaller than 0.1, making training unfeasible.
The possible solution is to modify current VQ encoder architecture and explore better training hyper-parameters. Or find a better dataset for training.
But I’m busy working, so … Hope anyone other could do that
I’m the one who create that fork. Yes, first train the encoder and then GPT. However, the current encoder loss cannot be optimized to be smaller than 0.1, making training unfeasible.
The possible solution is to modify current VQ encoder architecture and explore better training hyper-parameters. Or find a better dataset for training.
But I’m busy working, so … Hope anyone other could do that
@ain-soph Can you tell me how to fine-tune encoder and then GPT? Unfortunately I am not able to find script of finetuning anywhere. If you could guide me in this it would be great.
