基于机器学习算法的电子音乐信号辨识模型  

Research on the Discrimination of Electronic Music Signal Based on Machine Learning Algorithm

作　　者：杨文华[1] YANG Wenhua(Department of Normal,Weinan Vocational and Technical College,Weinan 714026,China)

机构地区：[1]渭南职业技术学院师范学院,陕西渭南714026

出　　处：《微型电脑应用》2021年第1期80-82,共3页Microcomputer Applications

基　　金：渭南职业技术学院2019年度项目(19WJYY08)。

摘　　要：电子音乐音乐信号辨识是一个种模式识别问题,当前音乐信号辨识方法存在误差大、速度慢,抗噪鲁棒性差等缺陷,为了提高音乐信号辨识正确率,提出了一种基于机器学习算法的电子音乐信号辨识方法。首先对电子音乐信号进行采集,并引入去噪方法对其进行预处理,抑制噪声对音乐信号辨识的干扰,提高抗噪鲁棒性,然后从去噪后的电子音乐信号中提取能够描述其类型的特征向量,其与电子音乐信号类型组成学习样本,通过采用机器学习算法的最小二乘支持向量机对学习样本进行训练构建电子音乐信号辨识模型,最后采用多种电子音乐信号进行验证性测试,结果表明,机器学习算法可以大幅度改善电子音乐信号辨识效果,可以提升电子音乐信号的识别速度,能够满足电子音乐信号的在线辨识,具有广泛的应用前景。The recognition of electronic music signal is a pattern recognition problem.At present,there are many defects in the recognition method of music signal,such as large error,slow speed,and poor noise robustness and so on.In order to improve the accuracy of the recognition of music signal,a recognition method of electronic music signal based on machine learning algorithm is proposed.Firstly,the electronic music signal is collected,and the denoising method is introduced to preprocess it,so as to suppress the interference of noise to the identification of music signal and improve the robustness of anti-noise.Then,the feature vector which can describe its type is extracted from the denoised electronic music signal,and the learning sample is composed of its type and electronic music signal type.The least square support vector machine is used to train the samples and to build the identification model of electronic music signals.Finally,a variety of electronic music signals are used for validation tests.The results show that machine learning algorithm can greatly improve the identification effect of electronic music signals,improve the recognition speed of electronic music signals,meet the online identification of electronic music signals,and have a wide range of application prospects.

关 键 词：电子音乐信号 机器学习算法 噪声干扰 辨识效果 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]