Regional dialect and talker sex information in high-frequency region

Abstract

High-pass filtering, in which the higher frequencies are retained and the lower frequencies are eliminated, has not been used in speech perception research as often as low-pass filtering because most intelligibility cues have been assumed to reside in the low-frequency region, up to 4 kHz (French and Steinberg, 1947). Recent research provided new evidence that listeners can also utilize high-frequency energy in the perception of voice, talker sex, and speech characteristics such as naturalness, pleasantness, and clarity, which improves speaker and word recognition in noise. Given the potential for the existence of accessible linguistic and paralinguistic information, the current study asks whether perception of regional dialect can be influenced by spectral content in the high-frequency region, and how robust this information is—in males and females—when low-frequency cues are unavailable. Listeners from Ohio heard phrases produced by males and females from Ohio and Western North Carolina high-pass filtered at 700, 1175, 1973, 3312, and 5560 Hz. Each higher filter provided increasingly fewer cues about dialect whereas sex identification remained relatively high. Female speech provided significantly more dialect cues than male speech when more spectral information was available (filter cut-offs at 1175 and 1973 Hz, but not at 700 Hz).