Model made by: @aiverse , Romance, Familia & Cinderella Soundtrack (13 mins) Pitch Extraction algorithm: RMVPE Epochs: 500 If you use please credit @aiverse on Youtube & Tiktok. DL:

Samples:

Can you please provide samples without effects?

Effects and reverb make too hard to see how your model really sounds.

yeah sure here:

Sounds good enough, altho i can hear some artifacting on her breaths.

yeah i probably overtrained it a bit, made this model awhile ago

Or you didn't clean the vocals enough.

Also, you must take a look at the Tensorboard.

nah the dataset was clean

yeah i always do

You did make sure to delete the inst bleed from the vocals?

yeah ofc, this isn't my first model lol

go