A METHOD FOR AUTOMATED SEGMENTATION OF SPEECH SIGNALS TO DETERMINE TEMPORAL PATTERNS OF NATURALLY EXPRESSED PSYCHO-EMOTIONAL STATES 

Alimuradov Alan Kazanferovich, candidate of technical sciences, director of student research and production business incubator, Penza State University (40 Krasnaya street, Penza, Russia), E-mail: alansapfir@yandex.ru
Tychkov Aleksandr Yur'evich, candidate of technical sciences, deputy director at the Research Institute for Basic and Applied Studies, Penza State University (40 Krasnaya street, Penza, Russia), E-mail: tychkov-a@mail.ru
Churakov Petr Pavlovich, doctor of technical sciences, professor, sub-department of information and measuring equipment and metrology, Penza State University (40 Krasnaya street, Penza, Russia), E-mail: churakov-pp@mail.ru 

004.934 

10.21685/2307-5538-2019-3-6 

Background. An assessment of human psycho-emotional state in the fields of professional activity associated with an increased risk of man-made and biogenic accidents is an important socially significant problem for the state. The aim of the study is to develop a method for automated segmentation of speech signals to improve the efficiency of determining temporal patterns of speech relevant to naturally expressed psycho-emotional states of a person.
Materials and methods. To develop the method, a unique technology for adaptive decomposition of non-stationary signals, namely, the improved ensemble empirical mode decomposition with adaptive noise, as well as the rule of differentiation based on the physiological aspect of speech formation, have been used. Software implementation of the method was performed in ©Matlab (MathWorks) mathematical modeling environment.
Results. A method for automated segmentation of speech signals into voiced, unvoiced, and pause sections to determine temporal patterns of speech reflecting naturally expressed human psycho-emotional states, has been developed. A study was conducted using a base of speech signals recorded from a group of subjects experiencing natural positive and negative emotions.
Conclusions. The results of the study have revealed that in conditions of instability of the speech apparatus motility, the developed method for segmentation makes it possible to more accurately determine the boundaries of voiced, unvoiced, and pause sections, thereby increasing the efficiency of calculating temporal patterns of speech, and determining psycho-emotional states of a person. 

speech signal processing, automation of processing, segmentation, adaptive decomposition, temporal speech patterns, naturally expressed psycho-emotional states