Listener preferences for alternative dynamic-range-compressed audio configurations

Some audio experts have proposed that using Dynamic Range Compression (DRC) to increase the loudness of music compromises audio quality. Conversely, in listening tests, researchers sometimes find that audio subjected to DRC is preferred over uncompressed audio. We test the hypothesis that it is DRC configuration, rather than the use of DRC per se, that determines listener preferences. In this study 130 listeners completed 13 A/B preference trials using pairs of RMS loudness-equalized stimuli subjected to different DRC configurations: viz., two magnitudes (heavy, moderate) and two compression types (limiting, compression) applied at three different points in the mix chain (track, subgroup, and master buss, here termed fullsum), along with an uncompressed control stimulus. Our results suggest that listeners prefer audio in which moderate compression has been applied to fewer signals simultaneously and dislike heavy limiting, particularly when applied to the full-sum, presumably because heavy DRC (and particularly limiting) applied to pre-mixed signals produces disagreeable distortion or because tracks whose amplitude characteristics would not have reached the DRC threshold alone may be deleteriously affected (e.g., attenuated) as a consequence of amplitude peaks in other tracks with which they are grouped.