Nikita Grebenyuk
Nikita Grebenyuk
Write this issue in that repo, not this (it is not related directly to esrgan).
> Yes. > > No, we provided the pre-trained model so you don't have to train it again by yourself. Please refer to our README about how to download those...
> Yeah I have encountered the same issues. The model just returns tens of "!!!!!!" and then cannot be conversed with anymore. This behaviour happens after the 4th round of...
For me it works when I have latest version of all packages, excluding torch (torch == 1.8.1+cu111, torchaudio==0.8.1, torchvision==0.9.1+cu111). But I am using this env only for inferensing/conversion. (training in...
You should have some file prepared where wavs paths are assigned with texts. Then run preprocess script as written in step 4 of "Pre-requisites" to convert text to IPA phonemes....
> Hello, > > I trained at 44KHz for a higher quality VC because the results were good when I trained with VCTK 22KHz. > > At this time, the...
Did you try **"fp16_run": false** ?
> You cannot train vits on a K80. K80's are a very weak GPU. You need at least a Tesla P100 or a T4 in order to avoid errors when...
Yes, it's quite slow, but results seems to be really the best from all open-source tts frameworks.
maybe problem in too new torch version, have you tried 1.8.1+cu111 ?