噪声环境中多轨道数字音频信号降噪方法  

Noise reduction method for multi⁃track digital audio signal in noisy environment

作　　者：赵丹 李蕊[1] ZHAO Dan;LI Rui(Guangdong Ocean University,Zhanjiang 524000,China)

机构地区：[1]广东海洋大学,广东湛江524000

出　　处：《现代电子技术》2024年第13期19-22,共4页Modern Electronics Technique

基　　金：湛江市哲学社会科学2022年度规划项目(ZJ22YB47);2023年广东海洋大学教学质量与教育改革工程项目:线上线下混合式课程钢琴(PX-112023073)。

摘　　要：为提升多轨道数字音频信号的峰值信噪比,降低噪声,使音频更加清晰,提出噪声环境中多轨道数字音频信号降噪方法。使用二进小波分解多轨道数字音频,将信号分解为不同的频率子带,使得噪声和信号在频率域上分离。通过模极大值计算分解信号的奇异性,由于信号和噪声在相同奇异性时的变化不同,因此能够确定信号中的噪声特点,利用层间相关搜索法找到分解后信号中的噪声并去除,最后使用交替投影法将去除噪声的分解信号重构,得到去除噪声的多轨道数字音频信号。实验结果表明:使用该方法进行降噪后,存在噪声的信号幅值得到了控制,一些单独突出可认定为噪声的信号基本消失;对低峰值信噪比的信号降噪,平均峰值信噪比提升了28 dB。A noise reduction method for multi-track digital audio signals in noisy environments is proposed to improve the peak signal-to-noise ratio(PSNR)of multi-track digital audio signals and reduce noise,so as to make the audio clearer.Dyadic wavelet decomposition is used for multi-track digital audio,and the signal is decomposed into different frequency sub bands to separate noise and signal in the frequency domain.The singularity of the decomposed signal is calculated by the modulus maximum.Because of the different variations of the signal and noise at the same singularity,the noise characteristics in the signal can be determined.The interlayer correlation search method is used to find out the noise in the decomposed signal and remove it.Finally,the alternating projection method is used to reconstruct the decomposed signal that has been denoised,so as to obtain a multi-track digital audio signal that has been denoised.The experimental results show that the amplitude of signals with noise is controlled after using this method for denoising,and some single prominent signals that can be identified as noise are eliminated basically.The average PSNR of the signal with low peak SNR has been increased by 28 dB.

关 键 词：多轨道 数字音频信号 信号降噪 二进小波 信号分解 模极大值 奇异性 交替投影法 

分 类 号：TN911-34[电子电信—通信与信息系统]