ASA 125th Meeting Ottawa 1993 May

2aSP27. Articulatory underspecification as a concept for controlling a speech synthesizer.

Bernd J. Kroger

Georg Heike

Claudia Opgen-Rhein

Inst. Phonetik, Univ. Koln, Greinstr. 2, 5000 Koln 41, Germany

A segmental model for generating articulator movements based on the principle of articulatory underspecification has been developed. This model is part of a computer-implemented speech production model controlling an articulatory speech synthesizer and capable of producing any utterance [B. J. Kroger, in Signal Processing VI: Theories and Applications, edited by J. Vandewalle et al. (Elsevier, New York, 1992), pp. 331--334]. Articulatory/phonatory control parameter time functions generated by this segmental model are compared to those generated by a gestural model. Qualitative comparison shows that the concept of segmental underspecification and the concept of cooccurrence (or overlap) of articulatory gestures result in similar articulator movements. [Work supported in part by DFG Grant No. He 434/21-1.]