VMD与PSO的乐器声音识别  

Recognition of Instruments' Sounds Based on VMD and PSO

作　　者：黄英来[1] 任田丽 赵鹏[1] HUANG Ying-lai;REN Tian-li;ZHAO Peng(Information and Computer Engineering Collager,Nortlieast Forestry University,Harbin 150040,China)

机构地区：[1]东北林业大学信息与计算机工程学院,黑龙江哈尔滨150040

出　　处：《哈尔滨理工大学学报》2018年第2期6-11,共6页Journal of Harbin University of Science and Technology

基　　金：新世纪优秀人才基金(NCET-12-0809);国家自然科学基金(31670717)

摘　　要：针对乐器音频信号的识别率低的问题,提出了一种变分模态分解(VMD)和被粒子群算法(PSO)优化的支持向量机(SVM)的乐器音频信号识别的方法。采用VMD将乐器音频信号分解成一系列平稳的窄带分量(IMF),并根据相关系数重构信号,采用小波去除残余的噪声。最后,在分析传统的声音特征提取方法基础上,提取梅尔频率倒谱系数(MFCC),用经PSO寻优参数的SVM进行音频信号的分类。实验结果表明,本文算法的去噪效果明显优于经验模态分解(EMD)和集合经验模态分解(EEMD)的分析结果;PSO优化后的SVM有效的提高了噪声环境下音频信号分类的正确率。Proposing the metliod that based on the variational mode decomposition(VMD)and particle swarm optimization(PSO)optimized support vector machine(SVM)are used to recognize the audio signals of tiie musical instruments aiming at the problem of the lowrecognition rate of musical instruments audio signals.In this paper,firstly,the instrument audio signals are decomposed into a series of stable narowband components(IM F)by VMD.After decomposition,according to the correlation coefficient we reconstruct the signals,then using the wavelet to remove the residual noises.Finally,based on the analysis of the traditional sound features extraction metiiod,extracting the Mel frequency cepstral coefficients(MFCC)and then SVM whose parameters are optimized by PSO is used to recognize the audio signals.This expserimental results show that the denoising effect of the proposed algorithm in this paper is better than that of empirical mode decomposition empirical mode decomposition(EEM D)%SVM optimized by PSO Ufectively improve the accuracy of audio signals’classification in noisy environment.

关 键 词：变分模态分解 小波去噪 梅尔频率倒谱系数 粒子群算法 支持向量机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]