摘要
为了解决传统电子音乐类型分类方法检测准确率不高的问题,提出一种多噪声背景的电子音乐类型分类方法。首先,降噪处理噪声环境下的电子音乐,得到含有噪声的电子音乐频谱,利用感知小波包变换提取噪声的特征,采样噪声的特征值并计算噪声的短时能量,得到电子音乐降噪表达式,计算频谱的二阶矩,得到不含噪声的电子音乐短时能量特征;然后,使用短时傅里叶变换计算得到每帧中所有电子音乐点的倒谱系数,利用倒谱系数得到计算电子音乐类型的二维矩阵,通过小波变换将噪声处理表达式与计算电子音乐类型的二维矩阵,处理为最终的噪声环境下的电子音乐类型检测表达式。实验结果表明,与两种传统噪声环境下的电子音乐类型分类方法相比,多噪声背景的电子音乐类型分类方法检测准确率更高,更适合在噪声环境下检测电子音乐类型。
In view of the low detection accuracy of the traditional electronic music type classification methods,a classification method of electronic music with multi⁃noise background is proposed.The electronic music in noisy environment is denoised to obtain the electronic music spectrum with noise.The perceptual wavelet packet transform is used to extract the characteristics of the noise.The eigenvalues of the noise are sampled and the short⁃term energy of the noise is calculated,so as to get the expression of electronic music denoising.The second moment of spectrum is calculated to get the short⁃term energy feature of electronic music without noise.Then the short⁃time Fourier transform is used to calculate the cepstrum coefficients of all electronic music points in each frame,and the obtained cepstrum coefficients are used to get the two⁃dimensional matrix for calculating electronic music types.The noise processing expression and the two⁃dimensional matrix for calculating electronic music types are processed into the electronic music type detection expression in the final noise environment by wavelet transform.The experimental results show that,in comparison with the two traditional electronic music type classification methods in noisy environment,the detection accuracy of the classification method of electronic music in multi⁃noise environment is higher,and it is more suitable for detecting electronic music type in noisy environment.
作者
孙刚平
SUN Gangping(Taiyuan University,Taiyuan 030012,China)
出处
《现代电子技术》
北大核心
2020年第21期110-113,共4页
Modern Electronics Technique
关键词
电子音乐分类
多种噪声环境
电子音乐降噪
噪声特征提取
倒谱系数
二维矩阵计算
electronic music classification
multi⁃noise environment
electronic music denoising
noise feature extraction
cepstrum coefficient
two⁃dimensional matrix calculation