ASA 129th Meeting - Washington, DC - 1995 May 30 .. Jun 06

1aSC31. Perception of voice-onset-time cued by articulatory information.

David R. Williams

Sensimetrics Corp., 26 Landsdowne St., Cambridge, MA 02139

This paper presents results of a first perceptual assessment of VOT continua synthesized using articulatory rather than acoustic parameters. The research employed Sensimetrics' HLSYN synthesis system, which embodies an articulo-acoustic model of speech production [K. N. Stevens and C. A. Bickley, J. Phon. 19, 161--174 (1991)]. The system permits time-varying control of fundamental frequency, vocal-tract shape, and of glottal, oral, and nasal orifice sizes; calculated aerodynamic quantities provide the basis for estimating KLSYN88 synthesizer source parameters. Six 10-member VOT continua (/ba/--/pa/, /da/--/ta/, /ga/--/ka/, /bi/--/pi/, /di/--/ti/, /gi/--/ki/) were constructed by varying the glottal adduction trajectory so that voicing onset occurred 0, 15, 30, ..., or 135 ms after oral release. In addition, two peak glottal openings (18, 30 mm[sup 2]) were tested; for /a/, rate of oral release was also varied (slow, fast). Subjects rated the ``goodness'' of the CVs as exemplars of the appropriate voiceless stop on a ten-point scale. In general, subjects preferred longer VOTs for velars than for labials. For alveolars, longer VOTs were preferred before /i/ than before /a/. Effects of peak glottal opening and oral release rate on VOT preferences were negligible. The results are discussed relative to traditional findings for acoustically defined VOT continua. [Work supported by NIMH.]