Speech Prosody 2002

Aix-en-Provence, France
April 11-13, 2002

Speaker-Ethnicity: Attributions Based on the Use of Prosodic Cues

Richard Todd

Speech and Hearing Research Group, Department of Computer Science, University of Sheffield, England, UK

The earlier work of [1] showed that listeners could discern speaker-ethnicity from both full and shortened stretches of casually-produced speech. The type of cues listeners used to accomplish this remained unclear, however. In this study the phonetic detail of speech was removed by filtering; both mono- and multilingual subjects were then required to determine the ethnicity of speakers using the prosodic cues that remained. Results show that attributions can still be made using such cues. Also seen was that monolingual listeners tended to be the better performers. Overall however, the task was less trivial than the majority of subjects had anticipated.

Reference

  1. Todd, R., 1998. Auditory Perception and Ethnic Group Attribution of Unknown Voices: Assessing the Robustness of Experienced Listeners’ Ratings when Confronted with Non-Native but Proficient English Speech. Proceedings of The Institute of Acoustics: Speech ‘98, 6, 343-350.

Full Paper

Bibliographic reference.  Todd, Richard (2002): "Speaker-ethnicity: attributions based on the use of prosodic cues", In SP-2002, 663-666.