Interaktivní systém pro detekci řečového signálu

Abstract

This thesis deals with the detection of speech and its processing in order to recognize the type of speaker (male, female, child), isolated word recognition, and an estimate of the vowels a, e, i, o, u. Methods of speech processing are carried out in time and frequency domain. Recognizing the type of speaker is based on finding fundamental period of vocal cords in cepstral area. To assist the recognition of isolated words (DTW algoritm), LPC coefficients are used, which are also used as constant parameters in the transfer function of the vocal tract in order to estimate the aforementioned vowels. Graphical user interface for the analysis of the audio recording is made in the computing program MATLAB. The outputs of each analysis are shown in graphical and text form.