Perceptual aspects of some voice qualities

Abstract

The perceptual dimensions related to the identification of four common, normal and distinct voice qualities were investigated. The qualities were representative of those heard (1) in everyday speaking (speech), (2) when someone produces a low-level whimper or cry (cry), (3) in country-western twang (twang), and (4) on the operatic stage (ring). Five subjects produced examples of these qualities at different fundamental frequencies. Thirty observers listened to tapes composed of combinations of the subject's paired stimuli and rated the degree of voice quality difference between the members of the pairs. The observers' dissimilarity matricies were then the input to INDSCAL. Depending on the subject who produced the simulus, three or four dimensions accounted for 43%–55% of the observers variance. In general, the most important dimension was Pitch followed by Nasality. Other dimensions included Brightness, Space, and Texture. Observers with different experimental background tended to weight the various dimensions somewhat experimental background tended to weight the various dimensions somewhat differently. The perceptual dimensions discovered help to explain the perceptual confusions that were noted when the same observers were asked to identify the same stimuli.