Level-matching before A/B comparison is required to evaluate processing objectively

The human auditory system consistently perceives louder sounds as ‘better’ — more open, more detailed, more exciting — independent of actual quality improvements. When comparing a processed signal against an unprocessed bypass, the processed version almost always has higher loudness (because limiting/compression raises the average level). Without matching gain before the comparison, the louder processed version will nearly always win, regardless of whether the processing actually improved the sound. Level-matching — typically by reducing the processed signal by the amount of gain added — makes the comparison fair and reveals the tonal and dynamic changes clearly. Most modern DAW processors include a ‘gain match’ or ‘auto-gain’ bypass for this reason.

Examples

Enable a limiter that adds 3 dB of average gain. In the limiter’s bypass, reduce the bypass level by -3 dB so both states play at the same perceived loudness. Now the A/B reveals only the tonal/dynamic differences the limiter introduces.

Assessment

You are comparing a compressed mix to an uncompressed mix and the compressed version sounds better. How do you know if the improvement is real or just louder? Describe the test procedure.