Loudness enhancement requires loudness-matched comparison to avoid bias toward the processed version

Increasing a mix’s loudness is achieved through a hierarchy of processing strategies, each with specific side effects: (1) Full-band ‘top-down’ squeeze — gentle low-ratio compression above a low threshold (approximately 3dB max gain reduction); side effects: upward expansion of low-level detail, reduced transient definition; (2) Full-band ‘bottom-up’ squeeze — upward compression targeting dynamic range below threshold; emphasizes background noise; (3) Band-limited limiting — targeting specific frequency regions. All strategies require level-matched comparison: comparing a louder processed version to a quieter unprocessed version makes the processing sound better than it is due to equal-loudness contour biases. Work simultaneously on two versions of the mix — one processed, one not — and match subjective loudness before evaluating quality.

Examples

Two stereo channels in the DAW: one raw mix, one passed through a limiter. Adjust the raw channel’s fader to match perceived loudness before toggling between them to evaluate the processing quality.

Assessment

You are comparing a heavily limited master to your unprocessed mix to decide if the limiting is hurting the sound. Why is this comparison inherently flawed unless you take a specific preparatory step? What is that step?