>>5893125
Vevo, made by the same people who made MaskGCT. It only needs 15-30 seconds of reference voice. Neat but has its drawbacks.
https://github.com/open-mmlab/Amphion/tree/main/models/vc/vevo

I made a version that integrates Ultimate Vocal Remover and auto segments anything over the max allowed length.
https://github.com/chameleon-ai/redubber