Speech Signal Reconstruction Based on Fractal Interpolation

Fractal theory is an active mathematic branch of nonlinear science. Since 1990's, it has been infiltrating into many f ields of information science, such as image processing, artificial intelligenc e and nonlinear circuitry and so on. As we know, time domain waveforms of speec h are very complicated and irregular, but they are similar in many details, and their contours have fractal features. Conventional speech processing methods bas e d on short term linear processing limit their power for greater achievement. In this paper, a fractal interpolative speech coding algorithm is applied to compr e ss and reconstruct speech waveform. Computer simulation shows that synthetical s p eech can be reconstructed with satisfactory quality by using this algorithm, mor eover the computational complexity is rather low.