From tongue model data to sound
Joan E. Miller, Osamu Fujimura · The Journal of the Acoustical Society of America · 1975
A simulation of the tongue, described in the preceding paper, “Computational Model of the Tongue” by Kiritani, Fujimura, Miyawaki, and Miller, provides three-dimensional coordinates of the nodal points of tetrahedra, forming a geometric shape which represents the shape of the tongue according to the specified muscular contractions. A computational procedure is described in which these data points are first approximated by a smooth function representing the tongue's surface subject to constraints imposed by assumptions about the surrounding rigid structure. Secondly, vocal-tract cross-sectional areas are estimated as differences between the assumed outer walls of the tract and the computed tongue surface. Finally, the formant frequencies may be inferred from the estimated area functions so that the resulting tongue positions may be evaluated by synthesizing a corresponding sound. Graphic displays illustrating the results are presented.