音乐情感自动分析研究  

Survey on music emotion automatic analysis

作　　者：蒋盛益[1] 李霞[1] 李碧[1] 王连喜[1] 

机构地区：[1]广东外语外贸大学信息学院,广东广州510420

出　　处：《计算机工程与设计》2010年第18期4112-4115,共4页Computer Engineering and Design

基　　金：国家自然科学基金项目(60673191);广东省自然科学基金项目(9151026005000002);广东省高等学校自然科学研究重点项目(06Z012)

摘　　要：音乐情感自动分析在音乐检索和音乐推荐等方面具有广泛的应用。对3种音乐情感模型进行了对比分析,介绍了音乐情感分类方法,并指出已有研究存在的不足。音乐分割与摘要是高效音乐浏览与推荐的基础,在对音乐分割与摘要方法进行分析的基础上,指出了定长分割策略的不足;借助音乐相似性与情感可视化实现音乐推荐,对音乐相似性度量与可视化方法进行了概述。最后,展望了对音乐情感自动分析的研究方向。Music emotion automatic analysis has wide range of applications in the field of music retrieval and music recommendation. Comparative analyses of three models of music emotion are given and classification methods to music emotion are introduced and then some shortage of research about music emotion automatic analysis are pointed out. Music segmentation and summarization are the basis of music browsing and music recommendation effectively. On the basis of analysis to music segmentation and summarization, the shortage of limit length segmentation method is pointed out. It is from music similarity and emotional visualization to complete music recommendation. A summarization is given to music similarity measure and visualization method. At last, the future of the research direction of music emotion automatic analysis is prospected.

关 键 词：音乐情感 情感识别 情感模型 音乐摘要 音乐检索 音乐推荐 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]