摘要
基于小波分解下的语音压缩编码与重构框架,研究分析了含噪情况下贪婪算法的重构性能和抗噪性能,提出了一种改进的自适应压缩采样匹配追踪算法(ACoSaMP)。该算法可在稀疏度未知的情况下,通过设置可变步长分阶段实现对稀疏度的逼近。同时,在每次迭代过程中,用最小二乘法对残差信号进行估计,代替传统CoSaMP算法对整个信号的估计。最后用小波去噪法对合成语音进行处理。实验结果表明:不同压缩比下,该算法的主客观重构效果均优于现有同类算法,对噪声有较强的鲁棒性。
This article ation based on speech analyzes greedy algorithmg reconstruction and anti-noise performances in noise situ- compression coding and reconstruction framework via wavelet decomposition. We propose an improved adaptive compressive sampling matching pursuit algorithm （ACoSaMP）. The algo- rithm can approximate the circumstances. Meanwhile sparsity in different phases by setting the variable step size in sparsity unknown , in every iteration, the residual signal is estimated using the least squares meth- od,instead of the traditional CoSaMP algorithm. synthesize speech. Experimental results show the outperform existing similar algorithms in different ness to noise. Finally, the wavelet de-noising method is employed to subjective and objective performances of the algorithm compression ratios and our algorithm has strong robut-
出处
《南京邮电大学学报（自然科学版）》
北大核心
2013年第1期16-22,共7页
Journal of Nanjing University of Posts and Telecommunications：Natural Science Edition
基金
国家自然科学基金(60971129)资助项目
关键词
压缩感知
重构算法
自适应压缩采样
匹配追踪
小波去噪
compressed sensing
reconstruction algorithm
adaptive compressive sampling
matchingpursuit
wavelet de-noising
