Speech and solo voices are the least revealing test sounds for loudspeaker flaws

Evidence

The most useful recordings had broadband, dense spectra, such as those shown in Figures 3.16 and 4.7. This confirms in a different way our observation over many years that if a loudspeaker has an audible problem, there is a high likelihood that resonances are involved. It also suggests a strategy for choosing program material: for demonstrations intended to impress, use simple sounds—solo voice, guitar, small combos and so on, especially with little reverberation—and, if possible, use a relatively dead room. To look for problems, listen to complex instrumental combinations with wide bandwidth and reverberation—and listen in a room with some reflections. — , Toole, Floyd E - Sound reproduction-…, 383

Categorization

• CANDY: A punchy, quotable inversion: the sounds we most care about (voices) are the worst for revealing equipment flaws, and the strategy for demoing vs. diagnosing gear flips the program choice.
• SEED: Angle: ‘why your expensive stereo review is based on the wrong songs’ — a piece on how demo music choice rigs both buying decisions and audio product reviews.
• XREF: connects to: Solo voice and simple instrument recordings least reveal loudspeaker flaws; Speaker cables and devices show near-zero audible differences in blind tests; Movie and broadcast speech is poor material for revealing loudspeaker flaws. (+2 more)