zhuangzhuangliu2345

Results 4 issues of zhuangzhuangliu2345

代码中好像只有one shot的train和inference,怎么进行zero shot ?

我注意到微调代码中语音采取的32k的代码进行load,为何有保存了32k的语音,保存的语音是用来做什么的?我想用16k的语音进行微调,不知道效果会不会有影响?

In follow-up

"I'm wondering if tuning multiple speakers simultaneously will make the tuned model adapt to multiple speakers or worsen it.