Zhuravlev V. Math modeling of speech production and noise-immunity conversion of natural origin acoustic signals.

The actual scientific problem of effectiveness increasing of the methods of processing of natural origin signals in informational systems with artificial and natural noise of the base of math modulating models of speech - hearing process is solved. Coding is made by the kinems of source speech system' articulation organs by means of modulation of statistic parameters of communication line substance energy: parameter of noise immunity of the message source, which is average in the time interval of phoneme signal, is sensitive to the base of kinem signal in the frequency band, which carries signal formants; the conversion of the energy amplitude of digital vortex of connection line matter according to the law of linear recurrence series determines parameters of potential noise immunity of message receiver and subjective noise immunity of message source. Energy, psycho-physical and informational adequacy of worked out math models is experimentally proved by the methods of temporal, correlation, statistic and spectral analysis of real speech signals phoneme parameters. Experimental statistic researches of noise immunity functions, which were undertaken, proved the adequacy of analog and digital models of speech messages source and receiver.