帧间自适应的压缩感知谱减去噪方法  

Speech denoising method based on adaptive inter-frame compressed sensing of spectrum subtraction

作　　者：畅江[1] 张雪英[1] 李凤莲[1] CHANG Jiang;ZHANG Xueying;LI Fenglian(Taiyuan University of Technology,Taiyuan 030024,China)

机构地区：[1]太原理工大学,山西太原030024

出　　处：《现代电子技术》2017年第9期14-17,共4页Modern Electronics Technique

基　　金：国家自然科学基金(61371193);山西省青年科技研究基金(2013021016-2);山西省回国留学人员科研资助项目(2013-034);山西省研究生创新项目(2015BY24)

摘　　要：针对应用压缩感知理论对含噪语音进行去噪其信噪比低的问题,以及应用谱减法对含噪语音去噪后语音信号仍不清晰的情况,提出帧间自适应的压缩感知谱减去噪方法。同时,由于传统压缩感知理论不能使语音信号在重构时实现帧间自适应的去噪效果,对此缺陷提出一种改进算法,并且将该算法应用到谱减法的去噪过程中。相比于经典的谱减法,实验结果表明,所提算法对含噪语音进行去噪不仅可以提高含噪语音的去噪效果,还可以有效地解决谱减法无法去除背景噪声及音乐噪声的问题。Since the noisy speech denoising based on compressed sensing theory has low signal-to-noise ratio,and thespectrum subtraction used to denoise the noisy speech still has poor speech signal,a speech denoising method based on adaptive inter-frame compressed sensing of spectrum subtraction is put forward.The traditional compressed sensing theory can′t realizethe adaptive inter-frame denoising effect while reconstructing the speech signal,so an improved algorithm is proposed,and applied to the denoising process of the spectrum subtraction.The experimental results show that,in comparison with the classical spectrum subtraction,the proposed algorithm used to denoise the noisy speech can improve the denoising effect of the noisy speech,and effectively eliminate the background noise and music noise that the spectrum subtraction can′t do.

关 键 词：压缩感知 谱减法 语音去噪 正交匹配追踪算法 

分 类 号：TN912.3-34[电子电信—通信与信息系统]