论文部分内容阅读
文章首先简单介绍了压缩感知(CS)理论框架,然后根据语音信号小波变换系数的特点,提出了改进的压缩感知算法,对高频系数进行压缩处理,低频系数不变。采用基追踪算法重构出高频系数,再利用小波反变换得到原始语音信号。实验结果表明,在相同的测量点数下,本文的算法比原有CS算法在重构语音的信噪比和MOS分上都有较大的提升。
In this paper, the theoretical framework of compressed sensing (CS) is briefly introduced. Then, based on the characteristics of wavelet transform coefficients of speech signal, an improved compressive sensing algorithm is proposed. The high frequency coefficients are compressed and the low frequency coefficients remain unchanged. Using the basis pursuit algorithm to reconstruct the high frequency coefficients, and then use the inverse wavelet transform to get the original speech signal. Experimental results show that under the same measurement points, the proposed algorithm improves the signal-noise ratio and MOS score of reconstructed speech more than the original CS algorithm.