How to plug-in new finetuned HiFiGAN?

bshall / knn-vc

Voice Conversion With Just Nearest Neighbors

https://bshall.github.io/knn-vc/

Other

450 stars 65 forks source link

How to plug-in new finetuned HiFiGAN? #34

Closed youssefabdelm closed 4 months ago

youssefabdelm commented 10 months ago

Hey! I know you wrote about this here: https://github.com/bshall/knn-vc/issues/23

I think I'm going to go ahead and try those steps now

However, I did find this 48K HiFiGAN model someone trained (seemingly I could be wrong): https://github.com/vtuber-plan/hifi-gan/releases/tag/v0.3.1

Is it possible to plug that .pt checkpoint into knn-vc as is or does it need to still be trained further?

EmreOzkose commented 9 months ago

I think your mentioned vocoder model is trained with mel-spectrograms, but this repo uses wavLM as audio encoder. Hence I think you should train again.