摘要
针对传统模型无法准确实现电子音乐分类和识别的难题,提出改进极限学习机的电子音乐分类模型。首先对电子音乐数据进行采集,并提取其倒谱系数特征,并采用核主成分分析对特征进行筛选;然后采用遗传算法对极限学习机的参数进行选择,并用于构建电子音乐的分类器;最后采用多种类型的电子音乐进行仿真实验,改进极限学习机的电子音乐平均分类率达到了95%以上,电子音乐的错分率要远远低于当前其他电子音乐分类模型。实验结果验证了该电子音乐分类模型的可行性以及优越性。
It is difficult to class and recognize the electronic music with the traditional model accurately, a new electronic music classification model based on improved extreme learning machine is proposed. The electronic music data is collected to ex- tract the feature of the cepstrum coefficient. The kernel principal component analysis is used to screen the feature. The genetic al- gorithm is used to select the parameters of the extreme learning machine to construct the classifier of the electronic music. The polytype electronic music is adopted to carry out the simulation experiments. The average classification rate of the electronic mu- sic can reach up to 95% with the improved extreme learning machine, and the wrong classification rate of the electronic music is far lower than that of other electronic music classification models. The feasibility and superiority of the electronic music classi- fication model were verified with the experimental results.
出处
《现代电子技术》
北大核心
2017年第5期155-158,共4页
Modern Electronics Technique
关键词
音乐分类
核主成分分析
极限学习机
音乐特征
遗传算法
music classification
kernel principal component analysis
extreme learning machine
music characteristic
ge- netic algorithm
