Yang Chengyue

Results 19 comments of Yang Chengyue

> 大佬, 按上面的步骤 1B-微调训练没有训练出来的模型文件 咋搞T-T 看看后台命令行

你填的list文件路径里好像有奇怪的字符(在D:\前面

> ![image](https://private-user-images.githubusercontent.com/86220968/301031946-d6883d93-e377-4d62-9fdf-a929476ff602.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.pKgkL9IOY6m7XHagkG18BtZhK__Qe4y5-w80lk9AdCI) 图上我看不出来,但是后台消息里显示有多一个字符

> > ![image](https://private-user-images.githubusercontent.com/86220968/301031946-d6883d93-e377-4d62-9fdf-a929476ff602.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.pKgkL9IOY6m7XHagkG18BtZhK__Qe4y5-w80lk9AdCI) > > 图上我看不出来,但是后台消息里显示有多一个字符 建议自行搜索\u202a,这个不是本代码库的问题或cpu训练的问题

> > > > > > 你s2_train.py怎么改的 > > 1. 注释了这一行 > > ``` > """Assume Single Node Multi GPUs Training Only""" > # assert torch.cuda.is_available() or torch.backends.mps.is_available(), "Only GPU...

最后报错的trainer不是s1_train里的trainer,感觉是pytorch-lightning这个库的问题,建议重新安装

似乎可以通过修改noise_scale来调整语速

> > 现在只能通过参考语音。后面想加语速滑竿的 > > 等待大佬出此功能!!! GPT-SoVITS\GPT_SoVITS\module\models.py里 类型SynthesizerTrn的decode方法 但是如此操作在调整过度的情况下会导致结果失真。