基于改进小波变换及神经网络的电子音乐信号识别方法  

Recognition Method of Electronic Music Signal Based on Improved Wavelet Transform and Neural Network

作　　者：张珺[1] 赵玉霞 ZHANG Jun;ZHAO Yu-xia(Shangluo Vocational and Technical College,Department of Normal Education,Shangluo 726000 China;School of Mathematics and Computer Application of Shangluo University,Shangluo 726000 China)

机构地区：[1]商洛职业技术学院师范教育系,陕西商洛726000 [2]商洛学院数学与计算机应用学院,陕西商洛726000

出　　处：《自动化技术与应用》2023年第5期48-51,共4页Techniques of Automation and Applications

基　　金：陕西省重点研发计划项目(2020GY-093);商洛市科技计划项目(SK2019-83)。

摘　　要：基于当前智能化发展水平,通过优化小波变换与神经网络,提出一种电子音乐信号自动化识别方法。合理扩大信号高、低频系数,经小波变换降噪,适当缩小高、低频系数,采用多输入多输出神经元,构建前向的径向基函数神经网络,将隐藏层神经元换成高斯激活函数,令信号呈中心径向对称形式,求解连接权重后,架构自动识别流程。仿真实验阶段,针对不同风格的电子音乐信号,检验该方法的降噪效果与自动识别准度、速度。实验结果验证出所提方法噪声滤除优势显著,且能够精准、快速地识别出目标电子音乐信号,具有较好的有效性与可行性。Based on the current level of intelligent development,an automatic recognition method of electronic music signals is proposed by optimizing wavelet transform and neural network.Reasonably expand the signal high and low frequency coefficients,reduce noise by wavelet transform,appropriately reduce the high and low frequency coefficients,use multiple input and multiple output neurons to construct a forward radial basis function neural network,and replace the hidden layer neurons with Gaussian activation functions,The signal is in the form of radial symmetry at the center,and after the connection weight is solved,the automatic identification process is constructed.In the simulation experiment stage,the noise reduction effect,automatic recognition accuracy and speed of the method are tested for different styles of electronic music signals.The experimental results verify that the proposed method has significant advantages in noise filtering and can accurately and quickly identify the target electronics.Music signal has better validity and feasibility.

关 键 词：改进小波变换 人工神经网络 改进神经网络 信号识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]