Standard studio calibration levels are measurably too loud for typical listeners

Evidence

Recommendation ITU-R BS.1116–2 (2014) chose to standardize the combined playback level from some number of active channels at 85 dBA, allowing 18 dB before digital clipping. Bech and Zacharov (2006) comment that they find this calibration level to be excessively high by 5–10 dB for subjects listening to typical program material. … So, the recommended calibrations for playback sound levels are frequently found to be excessively high, as judged by moviegoers and in private home theaters. The point of this discussion is that the choice of a playback sound level for product evaluation is a serious matter. It affects what can be heard, and how it is heard (see the equal-loudness contours in Section 9.4). Ideally, comparative listening should be done at a sound level that matters to you, or to the audience for your opinions. Olive et al. (2013) used program at an average level of 82 dBC, slow. This was for an audience of serious listeners doing foreground listening of a demanding kind. Most tests do not reveal their listening levels, so we never really can be certain what the results mean. — Toole, Floyd E, Sound reproduction- the acoustics and…, 1131

Categorization

• XREF: The warning that unpublished listening levels make test results hard to interpret echoes methodological skepticism familiar from other psychoacoustic testing literature.
• QUESTION: A real gap: most published listening tests omit the playback level, making cross-study comparison of perceived quality results unreliable.