Abstract: This paper presents an algorithm of speech enhancement based on two-dimensional and fractal theory. 2-D Fourier transform is done for the blocking and windowing of the speech data. After fractal dimension has been calculated, adapting filter based on the fractal dimension follows to reduce noise and enhance speech by data processing. Then the speech signal within the noise is reduced is obtained. The results of this study indicate that the fractal algorithm can remedy the shortage of conventional linear analysis method, 2-D adapting filter can make original character of speech signal be kept, and the algorithm reduces the unsteady noise and restrains musical noise effectiuely.