Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

About Different Size between Predicted-mel and Preprocess-mel #205

Open
ymzlygw opened this issue Aug 20, 2020 · 0 comments
Open

About Different Size between Predicted-mel and Preprocess-mel #205

ymzlygw opened this issue Aug 20, 2020 · 0 comments

Comments

@ymzlygw
Copy link

ymzlygw commented Aug 20, 2020

Hi,I am trying combine “deepvoice3-pytorch” with "wavenet_vocoder" ,which is both from your work, and I really thanks about that.

And , I extract the mel-output from deepvoice3-pytroch in synthesis.py of it on line 108 :

with torch.no_grad():
mel_outputs, linear_outputs, alignments, done = model(
sequence, text_positions=text_positions, speaker_ids=speaker_ids)
linear_output = linear_outputs[0].cpu().data.numpy()
spectrogram = audio._denormalize(linear_output)
alignment = alignments[0].cpu().data.numpy()
mel = mel_outputs[0].cpu().data.numpy()
mel = audio._denormalize(mel)

I save the mel output in .npy file and try to use it in wavenet_vocoder. But I meet the size-mismatch , while the preporcess-mel is (X,80) and can be used for synthesis to wave, but the predicted-mel from deepvoice3 is (80,X) and has size-mismatch error.

Firstly I think it maybe the transform problems so I change the predicted-mel(80,X) to .T(X,80) , but It didn't work too.

Could you please tell me why this happens? And how to modify the size of predicted-mel from deepvoice3 so that it match the input of wavenet?

I'm really want to know about it. Thanks.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant