A Survey of Multimodal Emotion Recognition

Abstract

Multi-modal emotion recognition refers to the method of recognizing human emotions in a more accurate way through the fusion of multiple signals, such as speech, text, expression, etc., which contributes more to the field of human-computer interaction and is also a more advanced direction in the field of artificial intelligence. The research of multimodal emotion recognition starts with the development process of speech emotion identification, combs and introduces the research situation of multimodal emotion recognition, introduces two emotion expression theories, and then introduces the relevant databases of emotion identification according to this classification. Then, the common methods of multimodal emotion recognition are analyzed from three dimensions: feature extraction, motif fusion and emotion classification. Finally, this paper summarizes the applications, challenges, and future directions of multimodal emotion recognition reference for other researchers in this field.