Efficient Bayesian Network for Music Style Classification
Mnsic style classification for commercial audience recommendation has become a promising application in computational auditory analysis.This fact raises the question of whether there is a classifier with SUfficient efficiency and simplicity to confront the great magnitude of music on the internet.In this paper with careful comparison with famous Neural Network classifier,we develop one easily-conceived and efficient algorithm to ruifill this request.In our method Naive Bayesian Network(NBN)is chosen as music style classifier.After examining and evaluating the conventional auditory features,we obtain eight important features to construct the specific NBN structure for this special classification,and find all the eight features could snfficiently determine the essential elements of music styles.The result from our NBN classifier,compared with that from Neural Network one.demonstrates that considerable improvement on efficiency and simplicity of music style classification could be achieved.
Classification Music Style Naive Bayesian Network
Yan Fu Liu Lin Junlin Zhou Li Rong Zhang Lin
国际会议
The International Conference Information Computing and Automation(2007国际信息计算与自动化会议)
成都
英文
1525-1528
2007-12-19(万方平台首次上网日期,不代表论文的发表时间)