Computation of mapping from muscular contraction patterns to formant patterns in vowel space
Yuki Kakita, Osamu Fujimura · The Journal of the Acoustical Society of America · 1983
In previous papers [J. Acoust. Soc. Am. Suppl. 1 62, S15(1977); in Frontiers of Speech Communication Research, edited by B. Lindblom and S. Ohman (Academic, London, 1979)], pp. 17–24, we demonstrated that a three-dimensional static model of the tongue can be used to explain some basic characteristics, in particular an insensitivity of the formant pattern of the vowel [i] to changes in the magnitude of muscle contraction force over a wide range when an appropriate set of muscles is used. We have implemented a computer program to compute systematically the mapping relation between various combinations of muscle contractions and the resultant formant frequencies in the vowel space. Muscular modeling for different vowels are being experimented based on available anatomical and physiological data, in particular those reported by K. Honda et al. [J. Acoust. Soc. Am. Suppl. 1 72, S103 (1982)]. Eventually, after the work by M. V. Mathews et al. [J. Acoust. Soc. Am. 63, 1535 (1978)] on mapping between area functions and formant patterns, this would illustrate inherent characteristics of the inverse relations from F patterns to muscular patterns for vowels. Uniqueness problems will be discussed with quantitative examples.