This can be used for voice conversion right?

[dillfrescott]

As in, train the vocoder just on one speaker, and when you inference, even with another speaker, it will sound like the one you trained on?

[dillfrescott]

Ok update, I trained it on a specific person and when I tried inferencing with an unseen person it sounded too much like the actual unseen person. I would have thought it would sound like the speaker it was trained on.

Would switching it to unconditional mode help?