摘要
音乐与情感有着非常密切的联系,发展针对音乐的情感识别系统,对于计算机音乐的研究与发展有着深远的意义.提出了一种基于PAD(Pleasure arousal dominance)模型以及基因表达式编程(GEP)算法的音乐情感自动识别方法.在众多音乐特征元素中抽取与情感关系密切的6个特征,并且采用PAD模型来描述音乐中的情感,在此基础上使用GEP算法实现对简单乐曲中单一情感的自动识别.从实验结果分析,本系统能够达到一个比较理想的识别效果和较低的识别误差.
Emotion has great impact on music composition.The development of emotion recognition systems for music is very important to the research of computer music.The paper presents a method based on pleasure arousal dominance（PAD） model and genetic expression programming（GEP） algorithm.In the paper,we advance 6 musical features which have close relations with music emotion and use PAD model to denote different kinds of music emotion.By using GEP algorithm,three formulae can be obtained by which the values of PADs in different songs can be calculated respectively.The experiment shows this method yields good results and achieves high accuracy.
出处
《厦门大学学报（自然科学版）》
CAS
CSCD
北大核心
2010年第6期798-802,共5页
Journal of Xiamen University：Natural Science
基金
国家自然科学基金资助项目(60975076)
