2pSC16. Timing of articulatory and acoustic targets---Evidence from electropalatography.

Session: Tuesday Afternoon, December 3


Author: Hector R. Javkin
Location: Speech Technol. Lab., Panasonic Technologies, Inc., 3888 State St., Santa Barbara, CA 93105
Author: Nicholas Kibre
Location: Speech Technol. Lab., Santa Barbara, CA 93105
Author: Karen Youdelman
Location: Lexington Ctr., Inc., Queens, NY


The synthesis of tongue-contact patterns analogous to those measured by electropalatography (EPG) has been undertaken both to provide models for the speech training of deaf children and to improve our understanding of articulatory gestures. [Javkin et al., ESCA Workshop on Speech & Language Technology for Disabled Persons, Stockholm, June 1993, 137--140; ICSLP S32-5.1 (1959--1962)]. In previous studies of other articulatory gestures [C. Browman, LabPhon3, UCLA, 1991], it was found that articulatory targets are not reached at the same time as the speech acoustic signal. In our present EPG study, a similar effect is found for tongue--palate contact---contact continues to increase after an acoustic target has been reached. For example, full frication for /s/ is achieved before the closing gesture is completed, and the decay of frication occurs after the release gesture begins. Even in consonants of relatively long duration (150 ms) from disyllabic words produced in isolation, there is, in effect, no steady-state contact---the release begins as soon as the closure is completed. This has led to adjustments in timing for palatographic synthesis derived from a standard text-to-speech system.

ASA 132nd meeting - Hawaii, December 1996