作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2012, Vol. 38 ›› Issue (08): 144-146. doi: 10.3969/j.issn.1000-3428.2012.08.047

• 人工智能及识别技术 • 上一篇    下一篇

音乐和歌词融合的歌曲情感分类研究

钟 将,程一峰   

  1. (重庆大学计算机学院,重庆 400044)
  • 收稿日期:2011-06-10 出版日期:2012-04-20 发布日期:2012-04-20
  • 作者简介:钟 将(1974-),男,副教授、博士,主研方向:文本分类,数据挖掘,知识管理;程一峰,硕士研究生
  • 基金资助:
    重庆市自然科学基金资助项目(CSTC2010BB2046);“211工程”三期建设基金资助项目(S-10218)

Research on Music Mood Classification Integrating Audio and Lyrics

ZHONG Jiang, CHENG Yi-feng   

  1. (College of Computer Science, Chongqing University, Chongqing 400044, China)
  • Received:2011-06-10 Online:2012-04-20 Published:2012-04-20

摘要: 为更好地对歌词进行情感分类,提出一种改进的基于类间差别的CHI特征选择方法。该方法可单独用于歌词情感特征提取,将选取的特征应用于支持向量机分类器中,融合音频特征与利用改进CHI方法选择的歌词特征对歌曲进行情感分类。实验结果表明,融合后的特征可以取得比任何单一种类特征更好的分类效果。

关键词: 情感模型, 歌曲情感分类, CHI统计方法, 支持向量机, 基于差别的CHI方法, 特征融合

Abstract: In view of the distinct levels of the association of a term with different classes, an improved difference-based CHI method is proposed to extract discriminative affective words form lyrics text. Support Vector Machine(SVM) classifier is constructed to apply the selected features, obviously increasing lyric sentiment classification performance. The lyric features selected by the improved method are combined with audio features. Experimental results verify the efficiency of this fusion.

Key words: mood model, music mood classification, CHI statistics method, Support Vector Machine(SVM), difference-based CHI method, feature fusion

中图分类号: