AL-Rafidain Journal of Computer Sciences and Mathematics

Abstract

In this research wavelet technique was used for speech compression. It was translated and scaled to groups which represent mother wavelet function.
Applying the adopted algorithm with MATLAB software using Daubechies filters family (with multi-scale for compression level) to compress speech signal for (N=1..9). When the compression level and the distortion level, a best compression level seen at level (5), with different type of wavelet and fixed compression ratio at level (5) with different type of Daubechies filters family.
Minimum signal distortion was seen when the adopted algorithm was applied. And with multi type of wavelet it was seen that the scale had direct proportional relation ship with the compression ratio, also speech compression does not need smoothing for the speech signal as compared to other compression methods.