Parametric description of the sound of speech in the task of recognition

Author(s):  A.V. Boldyshev, PJSC «Rostelecom», Belgorod, Russia, boldyshev@bsu.edu.ru

A.A. Medvedeva, Belgorod State University, Belgorod, Russia, medvedeva_aa@bsu.edu.ru

E.I. Prokhorenko, Belgorod State University, Belgorod, Russia, prokhorenko@bsu.edu.ru

Issue:  Volume 44, №23

Rubric:  Infocommunication technologies

Annotation:  The problem of recognition of speech signals, temporal, mel-cepstral and frequency characteristics of speech signals generated by various sounds of russian speech are analyzed. The complex approach of using various characteristics of speech sounds is proposed, which consists in the formation of groups and subgroups of sounds, depending on the value of the investigated parameters, on the basis of which a comparative analysis is carried out in an attempt to recognize a separate sound. The given approach can be used as an element of the recognition system, by means of which it is possible to limit the number of possible types of sound for further investigation

Keywords:  speech, speech recognition, sound characteristics

Full text (PDF):  Download

Downloads count:  380