Analýza řečových parametrů pro detekci klonovaného hlasu

Abstract

The thesis begins with an introduction to biometrics. It then discusses a theoretical investigation for human voice processing using MFCC features. It then discusses an introduction to speaker verification and identification systems. Next, the development of a speaker verification system using GMM and i-vector is described in detail. The practical part is devoted to the testing of the selected ASV systems and the evaluation of the tested parameters. The parameters tested are the cloning tools used, the gender of the cloned speaker, the type of microphone used for recording and external parameters such as recording length and noise.