I've been using Ilaria RVC for a long time, but sometimes the output has a weird issue: it has some rapid, extreme changes in volume, especially on long notes.
I'm pretty sure it's not the model cause 90% of the time outputs are okay.
This is what i've tried:
- WAV files vs. MP3 files: same issue
- Experimenting by changing the values for index influence, respiration filtering, envelope ratio and consonant breath protection: same issue
- Trying to upload same inputs with different volumes: same issue
- Trying to upload more compressed vocals: same
It's worth mentioning that the input files obviously didn't have this problem.
Any ideas?