problem
A steadier input lets the compressor work on tone instead of rescuing every phrase.
Quiet words may miss the compressor while loud words trigger heavy gain reduction, causing the vocal to lurch forward and backward.
The compressor then becomes a repair tool instead of a tone and density decision.
Use clip gain or pre-compressor automation to smooth phrase level, then meter how consistently the compressor responds.
The goal is not a flat vocal waveform, but fewer extreme jumps into the detector.
Raise quiet words, trim shouty notes, and let the compressor handle smaller movement after the vocal is already in range.
Meter Core helps compare vocal level, peak behavior, and final loudness while automation and compression are balanced.