A Humming Retrieval System Based on Music Fingerprint

摘要：

In this paper, we proposed an improved music information retrieval method utilizing the music fingerprint. The goal of this method is to represent the music with compressed musical information. Based on the selected MIDI files, which are generated automatically as our music target database, we evaluate the accuracy, effectiveness, and efficiency of this method. In this research we not only extract the feature sequence, which can represent the file effectively, from the query and melody database, but also make it possible for retrieving the results in an innovative way. We investigate on the influence of noise to the performance of our system. As experimental result shows, the retrieval accuracy arriving at up to91％ without noise is pretty well

关键词： Music Information Retrieval Music Fingerprint

作者: Xingkai Han Baiyu Cao

作者单位: Department of Computer Science and Technology arbin Institute of Technology Shenzhen Graduate Schoo Department of Computer Science and Technology Harbin Institute of Technology Shenzhen Graduate Schoo

会议类型: 国际会议

会议名称: 2010 International Conference on Signal and Information Processing(2010年IEEE信号与信息处理国际会议 ICSIP2010)

会议地点: 长沙

会议语种:英文

页码: 321-325

在线出版日期: 2010-12-14（万方平台首次上网日期，不代表论文的发表时间）