压缩感知新技术专题讲座(三) 第6讲 压缩感知在语音处理中的应用研究  

Compressive Sensing Theory and Its Application in Speech Signal Processing

作　　者：隋璐瑛[1] 黄建军[2] 张雄伟[3] 曾理[2] 王亚奇[4] 

机构地区：[1]解放军理工大学指挥自动化学院研究生1队 [2]解放军理工大学指挥自动化学院研究生2队 [3]解放军理工大学指挥自动化学院信息作战系 [4]中国人民解放军66072部队

出　　处：《军事通信技术》2012年第2期100-104,共5页Journal of Military Communications Technology

摘　　要：压缩感知CS(Compressive Sensing)作为一门新兴的技术,成为近年来人们广泛关注的研究热点。文中介绍了压缩感知理论的基本原理,在此基础上将压缩感知理论应用到语音信号处理中。首先研究了语音信号的稀疏性,说明了对语音信号进行压缩感知具有可行性;其次,采用随机滤波器组构造随机测量矩阵得到语音信号的压缩测量值;最后,研究了压缩测量值之间的相关性并将这种相关性作为稀疏度的一种度量方法用于控制随机滤波器阶数,实现了语音信号的自适应压缩感知。Compressive sensing is an emerging sampling theory proposed in recent years. The principle of CS was firstly introduced. Then the application of CS on speech signal processing was presented. Based on the speech signal＇s sparse representation, the speech signal was compressed using random filtering. The correlation of random measurements was also studied to allow adap- tive compressing. Finally the framework of adaptive speech signal compressive sensing was con- structed.

关 键 词：语音处理 压缩感知 稀疏表示 随机滤波 梯度投影稀疏重建 

分 类 号：TN912.3[电子电信—通信与信息系统]