ОБЗОР МЕТОДОВ КЛАССИФИКАЦИИ ЭМОЦИЙ НА ОСНОВЕ АНАЛИЗА АУДИОДАННЫХ С ПОМОЩЬЮ ГЛУБОКОГО ОБУЧЕНИЯ

Abstract

. In the past few years, there has been an increasing interest in the development of technologies aimed at determining the emotional state of a person. The most difficult in solving this problem is the subjectivity of emotions and the difficulty of tracking them. Deep learning has proven to be the most effective tool for emotion recognition, making it a particularly attractive area of research. In this article, we review recent research work on emotion classification, focusing on audio feature extraction and data augmentation techniques. We also compare and analyze existing CNN models to identify the most proven methods to best classify emotions based on audio data analysis. From the many publications devoted to the study of emotion recognition and developments in this area, we have selected only scientific articles, due to which the most relevant research is used in our article. At the same time, we consider only good quality articles that offer the best solutions to the research questions under consideration.