Responsible listening tests use discriminant-selected programs and within-program comparisons

Evidence

In these and all responsible listening tests the programs are chosen to reveal differences, not to disguise them (see Toole and Olive, 1988, and Olive, 1994). Several different recordings are used, averaging the results. Recordings with obvious spectral or bandwidth abnormalities are avoided so large program dependent effects are less likely. Also, loudspeaker comparisons are performed “within” individual programs, so although program is a variable, it is not one that is instantly compared. There will be some normalization of their effects during a comparison sequence. Listeners also get familiar with the program excerpts as they are repeated countless times throughout many listening tests. They become test signals—all pleasure from them is lost. — Toole, Floyd E, Sound reproduction- the acoustics and…, 550

Categorization

• DEFINE: Spells out the design criteria (‘reveal differences, not disguise them’) and why repeated familiarization matters — a clear methodology definition.
• SEED: Could support an essay on how repeated exposure converts musical pleasure into pure test signal — the cost of scientific rigor on listening.