Stelia I. Modeling human speech apparatus in problems of computer synthesis.

The present thesis is devoted to nomination for the degree of the Candidate in technical sciences, specialty 01.05.02 - mathematical modelling and computing methods. - Taras Shevchenko National University of Kyiv, Kyiv, 2010. This thesis is devoted to the computer reproduction of human voice on the basis of shared use of physical models of the voice source and vocal tract. The paper has further developed the methods of solving the issue of synthesizing the voice data, thereby solving the issue of artificial formation of sounds of a human voice based on the shared use of physical models of the voice source and vocal tract. There was developed an enhanced numeric algorithm of implementation of a two-mass model of the vocal folds of Ishizaka - Flanagan, built on the combination of a method for solving a system of equations of vibrations of two masses and a method of solving a nonlinear system for changes in pressure. On the basis of the developed algorithms there was created the program and algorithmic complex for studying of influence of various factors and parameters of the models in the language source - vocal tract system on speech production. For acoustical Klein - Gordon's equation the inverse problem of recovering the parameters of the vocal tract according to the measured output signal was solved.