170 epochs singing demo (some glitches, due to how the original vocals aren't 100% cleaned, and Ultimate Vocal Remover glitched up at one part of the original vocals):

170 epochs speaking demo (some glitches, probably due to how the original audio worked):

he sounds like moxxie/billy's va but if he was in an emo phase

Silver the Hedgehog (Sonic the Hedgehog, Bryce Papenbrook) - RVC v2 - 170/250 epochs

250 epochs singing demo (cleaner vocals used, but still not 100% clean):

250 epochs speaking demo (probably still has glitches):

using crepe instead of harvest on with songs can cause glitches.

Have you tried using harvest when doing the vocal replacement for the song?

This and the Blaze model ( will be the only models I trained with v2 and 40k sample rate. Future models will be trained with v3 and 48k (or perhaps higher) sample rate.

I will take that into consideration. Thank you.

170 epochs text-to-speech demo (using text from Sonic News Network's Silver the Hedgehog article):

250 epochs text-to-speech demo (I CANNOT get rid of the glitches without having the search feature ratio go too low):

170 epochs singing demo (copyrighted audio removed, some glitches, due to how the original vocals aren't 100% cleaned, and Ultimate Vocal Remover glitched up at one part of the original vocals):

250 epochs singing demo (copyrighted audio removed; cleaner vocals used, but still not 100% clean):

ITS NO USE

Wao