8
Speech Compression and Reconstruction Algorithm Based on Multi-Scale Wavelet Packet and Compressed Sensing
Authors: Jihua Cao, Xing Xiong, Jinpeng Yuan
Number of views: 464
Wavelet transform is widely applied in the field of the speech signal processing, but the wavelet decomposition
is only decomposing the low frequency coefficients, so higher resolution can not be obtained for the high
frequency components. In order to further decompose the high frequency coefficient, in this paper, the speech signal
was decomposed by multi-scale wavelet packet. By analyzing the sparsity of the decomposed coefficients, the
reconstruction algorithm based on compressed sensing was proposed by using the sparse representation of speech
signal, at the same time, the effect of compression ratio on the reconstruction quality was analyzed. In determining
the compression ratio, speech signal reconstruction quality and coding speed were taken into account, so the proposed
compressed and reconstruction method had good real-time property. A large number of simulation experiments
showed that the proposed method in this paper had excellent reconstruction quality and efficient coding efficiency.