Ability of normal hearing listeners to recognize vowels and musical instruments under spectrally-degraded conditions

Music perception requires greater spectral resolution than speech perception [Shannon, Int. Rev. Neurobiol. 70, 121 (2005)]. However, conclusions from these metadata are problematic given that they aggregate results from several different studies using diverse methodologies and paradigms. In particular, methodologies generally tapped into different perceptual dimensions, namely word recognition for speech and melody recognition for music. The present study aims to develop a paradigm to compare speech and music recognition based on similar perceptual dimensions, namely timbre. Stimuli consisted of naturally-spoken vowels and notes played on musical instruments as well as 32-, 8-, and 4- channel noise-vocoded versions. Listeners discriminated either instruments from vowels or vowels from instruments using a go/no-go task. The discrimination paradigm offered insight as to how available spectral information influenced perception between stimulus sound categories. Reaction times and accuracy were measured and organized as a function of signal degradation level to consider potential differences in how normal hearing listeners utilize spectral fine structure information when distinguishing vowels and musical instruments. In general, listeners’ ability to discriminate vowels from instruments under degraded conditions was similar to their ability to discriminate instruments from vowels. The results suggest that the perception of vowels and musical instruments rely on similar mechanisms.