DeEffect vocals by gilliaan

This vocal de-effecting and extraction model is designed to recover clean, dry lead vocals from processed recordings. It effectively removes reverb, delay/echo, vocal doubles, and backing-vocal bleed while simultaneously suppressing instrumental tracks. The model supports both stereo and mono effects. While primarily optimized for lead vocals, it can also process backing vocals if the channels are separated beforehand.

Essentially, the model integrates dereverberation, de-echoing, vocal de-bleeding, and vocal isolation techniques to extract a pure, underlying dry vocal signal.

Created by @gilliaan