In search of perceptual and acoustical correlates of polyphonic timbre

Polyphonic timbre refers to the overall timbre mixture of a music signal, or in simple words, the 'global sound' of any piece of music. It has been proven to be an important element for computational categorization according to genre, style, mood, and emotions, but its perceptual constituents have been less investigated. The aim of the study is to determine the most salient features of polyphonic timbre perception by investigating the descriptive auditory qualities of music and mapping acoustic features to these descriptors. Descriptors of monophonic timbre taken from previous literature were used as a starting point. Based on three pilot studies, eight scales were chosen for the actual experiment. Short musical excerpts from Indian popular music were rated on these scales. Relatively high agreement between the participants’ ratings was observed. A factor analysis of the scales suggested three perceptual dimensions. Acoustic descriptors were computationally extracted from each stimulus using signal processing and correlated with the perceptual dimensions. The present findings imply that there may be regularities and patterns in the way people perceive polyphonic timbre. Furthermore, most of the descriptors can be predicted relatively well by the acoustical features of the music. Finally the results suggest that spectrotemporal modulations are most relevant in the perception of polyphonic timbre