计算机工程 ›› 2007, Vol. 33 ›› Issue (02): 165-167.doi: 10.3969/j.issn.1000-3428.2007.02.058

• 人工智能及识别技术 • 上一篇    下一篇

基于音轨特征量的多音轨MIDI主旋律抽取方法

赵 芳,吴亚栋,宿继奎   

  1. (上海交通大学电子信息学院,上海 200030)
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2007-01-20 发布日期:2007-01-20

Melody Extraction Method from Polyphonic MIDI Based on Melodic Features

ZHAO Fang, WU Yadong, SU Jikui   

  1. (School of Electronic Information Engineering, Shanghai Jiaotong University, Shanghai 200030)
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-01-20 Published:2007-01-20

摘要: 在基于内容的数字音乐检索研究中,其音乐库大都直接使用复合音音乐数据文件。然而这种直接采用复合音音乐数据的索引必将给检索处理带来巨大的计算量以及复杂的匹配算法。该文提出了一种基于音轨特征量的多音轨MIDI主旋律信息音轨抽取方法。通过与人工标注结果的实验比较,表明该文实现的抽取方法可有效地从多音轨MIDI演奏数据文件中提取出主旋律音轨。

关键词: 哼唱检索, 复合音音乐, MIDI格式, 主旋律提取

Abstract: Much of the work on content based music query focuses on polyphonic music files. However the polyphonic format causes large calculation and complex algorithms for matching. This paper represents a model for auto melody extraction method for multi-track MIDI files based on melodic features. The results are compared with manual output and show that the implemented method can extract melody track from multi-track MIDI file effectively.

Key words: QBH, Polyphonic music, MIDI format, Melody extraction