1st ETRW on Speech Production Modeling: From Control Strategies to Acoustics 4th Speech Production Seminar: Models and Data Autrans, France May 20-24, 1996

A Fluid Flow Approach to Speech Generation

D. Sinder (1), G. Richard (1), H. Duncan (1), Q. Lin (1), James Flanagan (1), M. Krane (2), Stephen Levinson (2), D. Davis (3), S. Slimon (3)

(1) Rutgers University- CAEP Center, Piscataway, NJ, USA (2) Bell Laboratories, Murray Hill, NJ, USA (3) Electric Boat Corporation, Groton, CT, USA

A fluid dynamic formulation of speech generation may lead to an improved understanding of the physics of speech production. Unlike more traditional linear acoustic methods of speech synthesis, this alternate approach aims to capture more of the relevant physics by numerically solving a form of the Reynolds-Averaged Navier-Stokes equations describing fluid motion. Though computationally intensive, the method is not limited by assumptions of linearity and plane wave propagation inherent in linear acoustic analysis. Numerical simulations of flows in stylized vocal tract shapes, as well as measurements on physical flows are described. Special attention is given to fricative generation, since the physiological understanding, and subsequent synthesis, of these sounds stands to gain the most from this approach.

Full Paper

Bibliographic reference.  Sinder, D. / Richard, G. / Duncan, H. / Lin, Q. / Flanagan, James / Krane, M. / Levinson, Stephen / Davis, D. / Slimon, S. (1996): "A fluid flow approach to speech generation", In SPM-1996, 203-206.