Comprehensive Study of Automatic Speech Emotion Recognition Systems

Abstract

Speech emotion recognition (SER) is the technology that recognizes psychological characteristics and feelings from the speech signals through techniques and methodologies. SER is challenging because of more considerable variations in different languages arousal and valence levels. Various technical developments in artificial intelligence and signal processing methods have encouraged and made it possible to interpret emotions.SER plays a vital role in remote communication. This paper offers a recent survey of SER using machine learning (ML) and deep learning (DL)-based techniques. It focuses on the various feature representation and classification techniques used for SER. Further, it describes details about databases and evaluation metrics used for speech emotion recognition.