Abstract:

The work is devoted to the creation of models of blocks of various functional purposes of the speaker speech and voice analyzer. The purpose of the work is to simulate a language analyzer. The analysis of the physiological appearance of speech / voice, as well as the automated voice / speech recognition of the speaker has been carried out. For conducting experimental studies as a material, vowels were chosen because they are peculiar points of support in recognition of the language. On the basis of the conducted analysis, a functional model of the digital language analyzer, as well as a filter and block of emotion perception, including various types of lies, for this analyzer is constructed. The simulation of the recognition of changes in human psycho-emotional behavior at the biosignal when changing the voice/speech of the speaker is carried out. In the course of work, the model of the system of automatic recognition of vowels in the stream of the fused language is constructed that does not depend on the presence of noise in the speech signal and has the property of independence from the individual characteristics of the voice of the speaker. The automatic speech recognition system model is based on the acoustic analysis of the speech signal. The result of the periodic-aperiodic decomposition model with autoregressive parametrization and scaling factor is presented. The analysis of the choice of the digital analyzer, as well as the language analyzer filter is carried out.

Keywords: vocoder, voice, automated system