基于内容的音乐检索算法研究

Research of Content-based Music Retrieval Algorithm

基于内容的音乐检索把用户的哼唱作为输入,搜索更加智能,是目前音乐检索研究的热点和重点。在深入研究已有音乐检索算法的基础上,提出了基于基频矩阵的基音检测算法,有效提高了背景音乐下基音提取的准确性。对DTW算法进行了改进,引入高度动态调整和可变长搜索技术,提高了哼唱输入与音乐库中音乐匹配的性能。

Content-based music retrieval takes users＇ humming as inputs, which is more convenient and intelligible. Therefore, content-based music retrieval algorithm is full of interest to the researchers. After a deep research into currently available algorithms of content-based music retrieval, a new pitch detection algorithm based on the fundamental frequency matrix is proposed, which improves the accuracy of pitch detection under background music greatly. And the dynamic height adjustment and variable length searching are employed to modify the DTW （Dynamic Time Warping） algorithm. The modified DTW algorithm improves the matching performance between users＇ humming input and music in the database.