Content-based retrieval of music using mel frequency cepstral coefficient (MFCC)

Xin Luo, Xuezheng Liu, Ran Tao, Youqun Shi

School of Computer Science and Technology, Donghua University, Songjiang Districe, Shanghai, 201620, China

In the last few years, with the growing of multimedia in Internet, MP3 music become one of the most popular types. Some of the MP3 music collections available are approaching the scale of million tracks and this has posed a major challenge for searching, retrieving, and organizing music content. In this paper, we proposed a method to retrieve the MP3 lossy compression format music by using MFCC features. The Kullback-Leibler Divergence and Earth Mover’s Distance (EMD) are used to compute music similarity. Experiments show that the retrieving probability of our design can achieve high recall values of 95% out of a total of 1951 tracks in the database