Đỗ Trí Nhân

> I am facing same error in doing something similar Have you solve this problem? Yep, I have fixed this by install pytorch from source 

> Has anyone solved this issue? I also met this issue using Sagemaker container `763104351884.dkr.ecr.us-east-1.amazonaws.com/pytorch-training:1.10.0-gpu-py38-cu113-ubuntu20.04-sagemaker` Apex is not maintained for new version of CUDA, I think we should move to...

> Thanks for your reply! Indeed this has become more of a discussion topic than a problem. > As I mentioned the GPU memory usage is as expected, around 5-6GB...

> Dear contributors, > > I have applied G2PK, grapheme to phoneme conversion package, and achieved an improved Korean TTS results. > > This is the [link](https://joovvhan.github.io/glow-tts-demo-korean-mb-melgan-g2pk/) to the demo...

Me too, I train with only one speaker, it converged Then try to adapt for 65 speakers, 15 min each speaker I follow this  Finally I got Nan ...

> Have you fixed yet? @v-nhandt21 No, but I found that we can change the number of speaker from the pretrain, we just can replace the embedding layer of one...

> Sorry, I don't quite understand your question. Generally speaking, the length of speech that needs to be enhanced is generally the same as the input. If you want to...

@rohanbadlani @talipturkmen Have you try the soft DTW, did it converge? I found the Pseudocode in paper but, don't know how to use it with GPU

I guess that your melspectrogram generated by Tacotron is normalized, if it is true, you should de-normalize by mean and std.

> I run this, but get bad result, because there is noise in 4K frequency, I guess sub-stft-loss weight is too small, maybe need carefully adjust different loss weights I...