Research on Music Classification Based on MFCC and BP Neural Network
LiuYong Chun, Hong Song, Jing Bing Yang · Advances in intelligent systems research/Advances in Intelligent Systems Research · 2014
Because of the diversity and uncertainty of music, the classification rate and accuracy are both lower for the traditional classification methods in the large-scale music classification application.A based on BP neural network (BPNN) music classification method proposed in this paper can improve this performance, which extracts the feature parameters of music through mel frequency cepstrum coefficient(MFCC) firstly, and then the BPNN is used to train feature signals and establish the optimal classifier model, finally classifies the test music dataset.The average classification accuracy rate is up to 90.2%, and higher 7% than the HMM classification method by simulation experiments for the folk, classical, rock and pop different types of music, therefore, the results show that the BPNN is a quite effective music type classification method.