Music perception and speech perception have different sensitivities to acoustic variability

Word recognition is faster and/or more accurate when listening to a single talker than to multiple talkers in random orders. Subsequent research revealed that perception is challenged more by different-sounding talkers (in terms of voice pitch) than similar-sounding talkers. This experiment tested the extension of these findings to musical instrument perception. Listeners completed six blocks: three blocks of word recognition (by one talker, four similar-sounding talkers [similar voice pitch], four different-sounding talkers [varying voice pitch]) and three blocks of pitch labeling (by one instrument, four similar-sounding musical instruments [wind instruments with similar attacks and spectra], four different-sounding musical instruments [differing classifications, highly varying attacks and brightnesses]). Response times systematically increased and accuracy decreased as musical instrument variability increased. Response times slowed only for highly variable talkers compared to a single talker; word recognition accuracy was at ceiling throughout. Additionally, pitch labeling accuracy was positively correlated with measures of musical sophistication. Perception was overall sensitive to acoustic variability, which varied more across timbral dimensions for musical instruments (attack, spectral brightness) than for talkers (similar speech productions, some spectral differences across voices).