Rishikesh (ऋषिकेश)

Code is writing by making sure size should be 224 as mentioned in paper. In this line : https://github.com/rishikksh20/CeiT-pytorch/blob/d41cb23743fbb34c4354564a958f1d78de4e2770/module.py#L121 change w=7 and h=7 and do the same in this line...

@Oktai15 thanks !!!

ASR duration extractor and Gaussian Embedding seems very interesting.

@etimsijs Not in the case of Vocoder GAN like melgan, vocgan or TFGAN. In these GAN we take melspec as input and directly upsample to mels to hop size times...

yes I also saw that line but it won't impact my quality

hifigan v1 quality is better but hifigan is too slow.

yes better than hifi-gan with less training

I trained this model on my own dataset so I can't release pre-trained model but you can visit here : https://github.com/coqui-ai/TTS/pull/602 Coqui-ai TTS new release will contain this model which...

iSTFTNet

actually, it is a common occurrence when dealing with a variational autoencoder. Two way to resolve it 1) again start training from 3 or 4 back saved checkpoints (not from...