Návrh inteligentních metod pro kvantifikaci řečového signálu

Abstract

The diploma thesis deals with the analysis of the speech signal and the design of intelligent methods for the quantification of the speech signal. The proposed method is then tested to determine its success and use in speech signal classification. The theoretical part deals with the physical nature of sound; a theoretical description of biometrics; artificial intelligence and audio signal analysis. The practical part deals with the signal analysis in the time and frequency domain and the subsequent design of a classifier based on a convolutional neural network and its testing. In the final part of the thesis; the results of classification testing for a selected database of acoustic passwords are described; i.e. the total percentage success of the classification and the partial success for individual acoustic passwords.