基于小波包最优基的语音信号压缩方法

Speech Signal Compression Method Based on Wavelet Packet Best-Basis

提出了基于方差准则的小波包最优基压缩方法。首先选择一个合适的小波函数及分解层次对语音信号进行小波包分解。然后基于方差准则确定小波包最优基,保留语音信号的重要特征,实现信号压缩。最后对小波包最优基中的小波系数进行量化、编码,以便于信号在信道中传输。在接收端进行译码,重构压缩后的语音信号,观察压缩效果。实验中对语音信号采用不同方法进行了压缩实验,验证了本文算法的可行性。实验结果表明,本文算法的搜索量小,易于实现,压缩效果比较明显。

The compression method for the wavelet packet best-basis is presented based on variance criterion. Firstly,a suitable wavelet function and decomposition levels are selected and the wavelet packet decomposition is carried out for speech signal.Then,wavelet packet best-basis is determined based on the variance criterion,so that speech signal characteristics are retained and the signal is compressed.Finally,wavelet coefficients are quantized and encoded in the wavelet packet best-basis for transmitting the signal in the channel.At the receiving terminal,the signal is decoded and reconstructed for observing a compression effect.Different methods are used for speech signals to execute compression test.Experiments verify the feasibility of the algorithm.Results show that the algorithm is easy to be realized,and its search quantity is small.The compression effect of the algorithm is visible.