Pitch: 0 -- (Rap) Showcase #1: (Song - J Cole: Apparently) (Cleaned background vocals-No Reverb/Echo)

Pitch: -4 -- (Singing) Showcase #2 (Final): (Song - Camila Cabello: Havana) (Edited/No Reverb/Echo)

you should retrain with RMVPE

I don't use collab. When I tried using rmvpe, it went through 300 epochs in less than 5 minutes. Not sure, if that's normal. I'll try it later today

yeah it’s completely normal lol rmvpe is extremely fast

if it’s a good and clean dataset use mangio-crepe if not RMVPE is always the alternative to harvest

I'll try it out now then. I'll do up to 1000 just to test how it sounds EDIT - I tried infering after training was done. Keep getting errors when infering. Might have to wait.