Acoustic correlates of gender and ethnicity in speaker perception.

Abstract

Perceptual and acoustic explorations have determined that certain cues are evident in the auditory signal that provide information about speaker characteristics such as gender and ethnicity. Perceptual studies present stimuli to listeners who make decisions based on their perceptual interpretations. Results from such studies indicate that listeners are able to identify speaker ethnicity and gender [Lass et al., (1979); Boonstra et al., (2006)]. Acoustic studies analyze the speech to determine which cues, spectral or temporal, help listeners make their interpretations [Walton and Orlikoff, (1994); Xue and Fucci, (2001)]. The current study presents spectral and temporal analyses to provide a more comprehensive picture of various acoustic contributions to speaker voice types that listeners may utilize in their perceptual identifications. Speech samples (single word‐ and sentence‐length utterances) were produced by 20 male and female, African American and European American speakers of General American English. Results suggest that various spectral and temporal measures contribute to differences across voice characteristics. Differences attributed to speaker gender include fundamental frequency, intensity, pitch and amplitude perturbations, and the harmonics‐to‐noise ratio. Measures of stimulus intensity and pitch perturbation differed across speaker ethnicity. Differences resulting from the interaction of speaker gender and ethnicity include fundamental frequency and intensity. [Hope College Psychology Department, The Carl Frost Center for Social Science Research.]