Voice Processing and Voice-Identity Recognition

Abstract

The human voice is the most important sound source in our environment, not only because it produces speech, but also because it conveys information about the speaker. In many situations, listeners understand the speech message and recognize the speaker with minimal effort. Psychophysical studies have investigated which voice qualities (such as vocal timbre) distinguish speakers and allow listeners to recognize speakers. Glottal and vocal tract characteristics strongly influence perceived similarity between speakers and serve as cues for voice-identity recognition. However, the importance of a particular voice quality for voice-identity recognition depends on the speaker and the stimulus. Voice-identity recognition relies on a network of brain regions comprising a core system of auditory regions within the temporal lobe (including regions dedicated to processing glottal and vocal tract characteristics and regions that play more abstract roles) and an extended system of nonauditory regions representing information associated with specific voice identities (e.g., faces and names). This brain network is supported by early, direct connections between the core voice system and an analogous core face system. Precisely how all these brain regions work together to accomplish voice-identity recognition remains an open question; answering it will require rigorous testing of hypotheses derived from theoretical accounts of voice processing.