摘要
首先用离散傅立叶变换对音乐进行预处理,然后根据不同类型音乐的统计规律提取了特征向量.最后利用BP神经网络对特征向量进行训练,建立了一种基于神经网络的音乐分类方法.仿真实验结果表明,方法是一种有效的分类方法.
Firstly, the discrete Fourier transformation is used to the music, then the eigen- vectors of different types of music are found out according to the statistics law in this paper. Finally the eigenvectors are trained by BP neural network, a music classification method by neural network is presented. The simulations showed that the method is effective.
出处
《数学的实践与认识》
CSCD
北大核心
2014年第5期94-100,共7页
Mathematics in Practice and Theory
基金
广东省2013年本科质量工程项目
关键词
离散傅立叶变换
特征提取
神经网络
音乐分类
discrete Fourier transformation
feature extraction
neural network
music cat-egory
