betterftr

for me it generates samples 192x192 during training, trying to figure out why, since I set w and h 1024

Yes please!

modified line 64 in network/vocoders/hifigan.py from` device = self.device` to `device = torch.device("cuda")` now it says: AttributeError: 'HifiGAN' object has no attribute 'model'

Using this hifigan pretrained model from : `https://github.com/MoonInTheRiver/DiffSinger/releases/download/pretrain-model/model_ckpt_steps_1512000.ckpt` Can this cause issues?

any progress?

same for me, it was fine before yesterday

Re-download the application, extract it somewhere new, start it, window will pop up again that you allow or don't allow the connection, click yes, and its working again, then just...

I think I've found the cause for this issue, I deleted some models from the model_dir, cuz I trained an updated version of the voice, restarted the app then it...

For me this error occurred when I deleted some models, after restoring them from recycle bin the error stopped.

same