Yukong EN - Honkai: Star Rail 1.3 (RVC2) (100epoch) (9400steps) (RMVPE)

Anymore epoch it sounds like overtraining. the s sounds just get worse.

a model is also only as good as the inference audio

I think people over estimate how much training is needed.

this one is less quiet

omg im so glad someone else is also posting hsr models 🔥🔥🔥🔥

would it be rude to remake models for some of the characters you have already made? Mostly because I seem to have a bigger dataset and my models somehow sounds more natural with less epoch.

I'll probably finish making the ones that are missing first

i dont minf lol

that was when i didnt have the full dataset so i made do with whayi hsd

also because i was basing it off of arbitrary conclusions