adaptive_voice_conversion icon indicating copy to clipboard operation
adaptive_voice_conversion copied to clipboard

Published 20 hours ago • verified publisher icon jjery2243542

No training speaker encoder?

Open hyzhan opened this issue 5 years ago • 3 comments

Don't need to train a speaker encoder? There seems to be no such step in the code. Do we need to use a pre-trained model?

hyzhan avatar Sep 10 '19 03:09 hyzhan

Hi, speaker encoder is trained from scratch.

jjery2243542 avatar Sep 10 '19 04:09 jjery2243542

@jjery2243542 Is it enough to use only the reconstruction loss of the audio, does not need classification loss?

hyzhan avatar Sep 10 '19 04:09 hyzhan

Yes. only reconstruction loss.

jjery2243542 avatar Sep 10 '19 06:09 jjery2243542