A survey on music emotion recognition using learning models

Yixin Wang, Xujian Zhao, Chuanpeng Deng, Yao Xiao, Haoxin Ruan, Peiquan Jin, Xuebo Cai

Abstract: Music is one of the art forms for expressing emotions, and it conveys emotional information through elements such as combinations of notes, melodic variations, rhythmic modulation, and choice of timbre. Recently, music emotion recognition has attracted the attention of researchers due to that it can be widely applied under different scenarios, such as music recommendation systems, intelligent music generation and creation, music therapy and emotion regulation, and other fields. With the development of Artificial Intelligence, deep learning-based music emotion recognition technology has gradually replaced traditional machine learning technology, becoming the mainstream of the times. The purpose of this paper is to present a summary of current studies on music emotion recognition. Firstly, we introduce the task of music emotion recognition from some relevant definitions, processes, and emotion models. Then, the main current advances in music emotion recognition are detailed in terms of both traditional machine learning and deep learning. Next, some commonly used public datasets are presented, as well as evaluation metrics. Finally, we summarize the current research challenges facing music emotion recognition and the future trends.

External IDs:dblp:journals/mms/WangZDXRJC25