摘要
压缩感知是一种结合采样和压缩的新技术,是近年来研究的热点。文中研究基于压缩感知(CompressedSensing,CS)理论的语音信号处理新技术。验证了语音信号在离散余弦变换域(Discrete Cosing Transform,DCT)的近似稀疏性。根据文献[1]提出的最优观测理论,文中针对语音信号进行了研究,提出一种语音信号的最优观测矩阵算法。结合语音信号的近似稀疏性与最优观测矩阵算法,提出了基于最优观测的语音信号CS方法。实验研究语音信号在DCT域的CS重构性能。实验表明,基于最优观测的语音信号CS性能较好,验证了文献[1]理论的正确性。
Compressed sensing（CS） which offers a joint sampling and compression processes is a research hotspot in recent years. This paper researchs a new processing technology of speech signal based on CS. The approximate sparsity of speech signal in the DCT domain is verified. Based on the theory in the litera- ture [ 1 ] , An optimal observation matrix algorithm for speech signal is intruduced. Speech compressed sensing based on optimized observation is proposed on the combination of the approximate sparsity of speech signal and the optimized observation matrix algorithm. The performance of speech CS in the DCT domain is analyzed by the experiments. The experiments＇ results show that performance of the speech signal CS based on the optimal observation is better than which of other algorithmes. The results verifys the correctness of the theory in the literature [ 1 ].
出处
《南京邮电大学学报（自然科学版）》
2011年第6期49-54,共6页
Journal of Nanjing University of Posts and Telecommunications：Natural Science Edition
基金
国家自然科学基金(60971129)
国家重点基础研究发展计划(973计划)(2011CB302903)资助项目
关键词
压缩感知
离散余弦变换
稀疏性
内聚值
最优观测矩阵
正交匹配追踪
compressed sensing
discrete cosing transform
sparsity
mutual coherence
optimal observation matrix
orthogonal matching pursuit
