quickvc / QuickVC-VoiceConversion

QuickVC: Any-to-many Voice Conversion Using Inverse Short-time Fourier Transform for Faster Conversion
MIT License
225 stars 26 forks source link

Training recipe #16

Closed avi33 closed 1 year ago

avi33 commented 1 year ago

In section 2.3.2 in paper, you have mentioned that the input in speaker encoder varies between training stages. Can you please point ib code where it takes place?

Do yiu train end to end or given trained generator? Thanks in advance

guoyingying432 commented 1 year ago

Please refer to this: https://github.com/quickvc/QuickVC-VoiceConversion/issues/10

guoyingying432 commented 1 year ago

the model is trained end to end

avi33 commented 1 year ago

Thank you for your quick response.