First: have you read our voice model conversion guide? This is the definitive source for the best way to prepare your audio for voice conversion. If you've read this guide and your input audio meets these standards: carry on!
If your converted audio sounds thin or scratchy, especially on higher notes, it’s likely that your input audio exceeds the pitch range of the model. Try transposing your input audio down -12 semitones (one octave) in the advanced settings dropdown or select a different model with a higher range.
To ensure the highest quality audio conversions, make sure your input audio has a strong fundamental frequency (make sure you weren’t recording too far away from the mic).
The quality of your input audio is key for AI conversions. Any input audio with reverb, harmonies, or other effects may also hinder the pitch detection and conversion of the voice models. We recommend tidying audio with our Vocal Separator or finding a cleaner audio file to convert before proceeding.