XXXXRT666
XXXXRT666
Cause `curl` or `wget` doesn’t produce inference logs. Are you seriously under the impression that I’m explaining how to invoke an API to you? You’re setting up your own TTS...
Is the inference being performed on the CPU? If that’s the case, much more slower performance is expected. Based on my tests, it’s approximately 70 times slower.
use faiss-cpu from anaconda instead of pip
list路径
> 路径是对的,里面也有list,我放在D盘一直提示没权限,这怎么搞? 你的路径从来就没对过,一个文件夹还指望有多少权限?
把pickle删了
检查代码,看看import的是哪里的CNHubert,你import的可能是Main分支
> 我clone的应该是Main分支。 api_v2.py和V3主要特点是什么，我还不明白。是不是现在流行用 fast_inference_ 分支的api_v2.py、v3? 去看看`cnhubert.__file__`
> TTS.py 的一部分引用 import之后让他print一下看看`__path__`
记错了，是`__file__`