The Psychoacoustic Clipper clips the signal while avoiding sizzling or crackling distortion. The end result sounds somewhere between a limiter and a normal clipper.
The clipper uses block-based FFT filtering to remove distortion that isn't masked by the sounds in the input.
Because the filtering introduces overshoots, the clipper repeats the clipping-filtering process a few times to achieve peak level control.
Sometimes clipping sounds fine. Sometimes it sounds horrible. The difference is in the spectral content of the sound being clipped.
Sounds that have sparse spectral content (i.e. solo piano or vocal) can expose even just a little bit of clipping.
Sounds that have dense spectral content (i.e. snare drum) tend to hide clipping distortion well.
The psy-clipper compares the spectrum of the input and the spectrum of the distortion, and removes the part of the distortion that would "stick out". Example with piano:
original | normal clipper | psy-clipper |
![]() |
![]() |
![]() |