Systém pro analýzu řeči - identifikace řečníka

Abstract

This bachelor thesis addresses the issue of automatic speaker identification by voice. It summarizes the basic methods of speech signal analysis. There is a brief description of the principle of creating a human voice and its modeling. Furthermore, the tasks of the speaker recognition system and the identification method are classified. The main goal of this work was to design a software solution for the identification system. The system is designed in the MATLAB environment. The system uses melcepstral coefficients and the basic vocal cord tone period to identify the speaker. An artificial neural network is used for classification. A description of the algorithm of identification and evaluation of the success rate are part of this work.