语音信号幅值分布的统计分析

Statistics and Analysis of Speech Signal Amplitude Distribution

针对PCM编码的语音信号,提出了一种基于正交矩阵变换的语音处理方法以改变语音信号幅值的分布规律,从而减小语音信号的动态范围与标准差。实验表明,正交矩阵变换算法可以明显降低语音信号的幅值范围,实验数据显示变换后语音信号的标准差降低20%。同时多人试听实验证明解码后的语音质量没有降低。此算法缩小了语音信号的幅值范围和标准差,从而可以用更少的编码比特完成语音信号的编码,压缩效率可以进一步提高。

In view of PCM（Pulse Code Modulation） speech signal, this paper puts forward a method of speech processing based on orthogonal matrix transformation to change the amplitude distribution of speech signal, which can reduce the standard deviation of speech signal. Experiments indicate that the orthogonal matrix transformation algorithm can obviously reduce the amplitude range of speech signal, and the results show that and speech signal standard deviation is reduced by 20% through the transformation. At the same time, voice quality after decoding is not decreased according to listening experimenter. The algorithm reduces amplitude range and standard deviation of the speech signal, so speech signal can be coded with less bits and compression efficiency can be further improved.