Load/save voice profile?

[mindplay-dk]

Is there any way to save/load a voice clone?

I don't know in practice how much overhead it is to voice clone from a .wav file every time? I'd imagine it's quite a bit of work?

I'm thinking about building a fully free and open-source Spanish course, and I would need to generate thousands of clips with the same 5-6 voices.

This project looks amazing btw! Looking forward to play around with this. 😄

[ServeurpersoCom]

I plan to save the latents in omnivoice.cpp and qwentts.cpp, as with acestep.cpp
like a --ref-latents or --ref-rvq to save on wav encode

[ServeurpersoCom]

Done, you can produce the latents with the codec tool and reuse it at will to skip the encoding step.