对多声道音频文件的基于旋律的音频检索系统设计

Design of multi-channel audio retrieval system based on melody

基于音频相关属性信息进行检索的音乐检索系统具有较大的局限。近年来,基于旋律的音频检索技术的研究成为音频检索技术的热点课题,对单声道音频的旋律特征提取算法作了较深入的研究。本文在分析音乐旋律表达的基础上,研究多声道文件的旋律特征,提出了多声道文件的旋律提取方法,设计完成了面向多声道MIDI格式文件的基于哼唱的音频检索系统。结果表明,系统具有良好的检索功能。

The research has been focused on melody-based music retrieval technology because of the limitation of the music retrieval system based on the relative audio property. Much attention has been paid on the arithmetic of single-channel audio＇s melody feature extraction. This paper discussed the music melody representation and multi-channel audio＇s melody feature. The method of multi-channel melody extraction and a query by humming （QBH） based audio retrieval system was set up. The experiment showed that the retrieval system worked well.