Computer Engineering and Applications ›› 2022, Vol. 58 ›› Issue (4): 64-72.DOI: 10.3778/j.issn.1002-8331.2108-0538

• Research Hotspots and Reviews • Previous Articles     Next Articles

Survey of Music Emotion Recognition

KANG Jian, WANG Hailong, SU Guibin, LIU Lin   

  1. College of Computer Science and Technology, Inner Mongolia Normal University, Hohhot 010000, China
  • Online:2022-02-15 Published:2022-02-15

音乐情感识别研究综述

康健,王海龙,苏贵斌,柳林   

  1. 内蒙古师范大学 计算机科学技术学院,呼和浩特 010000

Abstract: Music is an important carrier to express emotion, and music emotion recognition is widely used in various fields. In the current research on music emotion, there are many problems such as the scarcity of music emotion datasets, the difficulty of emotion quantification, the limited accuracy of emotion recognition, etc. How to use artificial intelligence method in the effective and high-quality identification of music emotion trend has become the hot spot and difficulty of current research. This paper summarizes the present research status of music emotion recognition from music emotion datasets, music emotion model, music emotion classification method three aspects to comb, enumerates the current publicly available datasets and carries on the brief summary, comprehensively evaluates the common music emotion model, summarizes different classification methods according to different modals. Finally, the current problems and future research work in this field are summarized to provide ideas for further research.

Key words: music emotion recognition, music emotion model, music emotion classification, music emotion dataset

摘要: 音乐是表达情感的重要载体,音乐情感识别广泛应用于各个领域。当前音乐情感研究中,存在音乐情感数据集稀缺、情感量化难度大、情感识别精准度有限等诸多问题,如何借助人工智能方法对音乐的情感趋向进行有效的、高质量的识别成为当前研究的热点与难点。总结目前音乐情感识别的研究现状,从音乐情感数据集、音乐情感模型、音乐情感分类方法三方面进行梳理,列举当前可使用的公开数据集并对其进行简要概括,综合评判常见的音乐情感模型,针对不同模态总结不同的分类方法。最后对该领域当前问题及今后研究工作进行归纳概括,为后续进一步的研究提供思路。

关键词: 音乐情感识别, 音乐情感模型, 音乐情感分类, 音乐情感数据集