DCT域的语音信号自适应压缩感知 .pdfVIP

学兔兔 第31卷 第6期 仪 器 仪 表 学 报 V01．31 No．6 2010年6月 Chinese Journal of Scientific Instrument Jun．2010 DCT域的语音信号自适应压缩感知米 郭海燕 ，王天荆 ，杨 震 (1 南京邮电大学信号处理与传输研究院 南京 210003；2 南京工业大学理学院 南京 210009) 摘 要：利用语音信号在DCT域的稀疏特性，提出了一种基于语音信号能量特性的帧间与帧内自适应压缩感知方法。首先根 据语音信号各帧能量和各帧所属能量区间自适应分配每帧观测个数，称之为帧问自适应。然后当帧观测个数确定后，根据每帧 完全观测各分量的能量自适应选取相应个数的观测，称之为帧内自适应。实验表明，用该算法重构的语音具有较高的平均分段 信噪比和平均意见得分。 关键词：压缩感知；语音合成；稀疏性；L1优化；自适应 中图分类号：TN912 文献标识码：A 国家标准学科分类代码：510．40 Adaptive speech compressed sensing in the DCT domain Guo Haiyan ，Wang Tianjing ，Yang Zhen ( Institute ofSignal Processing and Transmission， ng University ofPosts and Telecommunications，Nanfing 210003，China； 2 College ofScience，Na ng UniversityofTechnology，Na ng 210009，China) Abstract：Utilizing the sparsity of speech signals in the DCT domain，this paper proposes an inter—frame and intra— frame adaptive speech compressed sensing(CS)method based on speech energy．Firstly，the inter—frame adaptive algorithm distributes the measurement number of each frame according to the energy of each frame and the energy in- terval each frame belongs to．Secondly，the intra—frame adaptive algorithm generates the number of measurements of each frame according to element energy of the corresponding frame full random measurement．Simulation results dem- onstrate that the speech recovered using the inter--frame and intra—-frame adaptive speech CS method has good average segment signal to noise ratio and mean opinion score． Key words：compressed sensing；speech synthesis；sparsity；L1 optimization；adaptive