The architecture of the emotion recognition program by speech segments

Abstract

The article discusses the issues of automatic recognition of the emotional coloring of a person's utterance using machine learning methods. In the software being developed, a change in the basic tone of a person's voice is taken as a basis. The architecture of the program is based on a multilayer perceptron. The result of the classifier is the choice of one of eight possible classes of emotions. In the course of the study, the accuracy of the classification of emotions was determined and the main directions of the development of the program were outlined.