Perceptual (but not acoustic) features predict singing voice preferences

Abstract

Singing is ubiquitous and highly relevant to human experience, but what governs our singing voice preferences is not understood. We take an empirical approach to determine the extent to which listeners’ liking of singing performances can be traced back to objective attributes of the voice stimuli. We show that which singing voices are liked cannot be explained based on purely acoustic features of the voice signal. Instead, perceptual features (partially) ground singing voice preferences, with a significant proportion of the variance in liking ratings explained based on perceptual features, such as (perceived) pitch accuracy, resonance, and articulation of singing performances.