Possible to produce good quality unconditional speech?

r9y9 / wavenet_vocoder

WaveNet vocoder

https://r9y9.github.io/wavenet_vocoder/

Other

2.33k stars 500 forks source link

Possible to produce good quality unconditional speech? #138

Closed realtimecrime closed 5 years ago

realtimecrime commented 5 years ago

Hi, many thanks for this great project. I am trying to reproduce unconditional speech samples. Is it possible to achieve good sound quality (comparable to the samples provided by google's wavenet) without conditioning on mel spectrograms with this code? I am training a model without local and global conditioning right now, but so far (160k steps) it only produces noise.

r9y9 commented 5 years ago

Not sure if it helps, but you can find some samples at https://github.com/r9y9/wavenet_vocoder/issues/1#issuecomment-354586299.

stale[bot] commented 5 years ago

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.