Lip2Wav icon indicating copy to clipboard operation
Lip2Wav copied to clipboard

Published 20 hours ago • verified publisher icon Rudrabha

Results of training on my own dataset

Open dongdongdashen opened this issue 3 years ago • 5 comments

Hi,When I train with my own dataset, the result is as shown in the following picture.Can you tell me where is the problem? May be the length of data is too short? (my dataset is about 20 minutes ) Looking forward to your reply ： step-5000-align step-5000-mel-spectrogram ）

dongdongdashen avatar Feb 28 '22 08:02 dongdongdashen

Hi, is it possible you've got teacher forcing enabled? Going by the attention matrix, the encoder doesn't seem to have an impact on the decoder

Domhnall-Liopa avatar Mar 02 '22 10:03 Domhnall-Liopa

Hi, is it possible you've got teacher forcing enabled? Going by the attention matrix, the encoder doesn't seem to have an impact on the decoder

Hi,I check the hparams.py and train.py in synthesizer folder,how can I stop teacher forcing when training? is it ok? 0N%31G23$ZPP)( ROOA4E0H

dongdongdashen avatar Mar 02 '22 11:03 dongdongdashen

I think if you use the following in hparams.py:

tacotron_teacher_forcing_mode="constant",
tacotron_teacher_forcing_ratio=0.,

It will disable teacher forcing in training

Domhnall-Liopa avatar Mar 02 '22 11:03 Domhnall-Liopa

I think if you use the following in hparams.py:

tacotron_teacher_forcing_mode="constant",
tacotron_teacher_forcing_ratio=0.,

It will disable teacher forcing in training

thanks!

dongdongdashen avatar Mar 02 '22 12:03 dongdongdashen

Could you share the attention plot after disabling the teacher-forcing? Did you solve the problem?

comvee avatar Mar 10 '22 06:03 comvee