Fractal properties of speech signals
Yu.V. Andreyev, A.S. Dmitriev
Institute of Radio Engineering and Electronics
of the Russian Academy
of Sciences, Moscow, Russia
It is known that digitized speech is widely used in modern communication systems, e.g., digital communication lines now begin to substitute ordinary analog telephony. And though a number of efficient methods of digital sound compression is developed by far, there is still a need in new efficient methods. In this report we study a possibility of application of fractal methods to compression of speech signals.
The method of fractal interpolation of functions is used, that is, a digitized speech signal waveform is approximated with a piecewise self-affine fractal interpolation function. In this method, a speech frame is divided into larger pieces called domains, and smaller ones called regions, and an optimal set of affine transformations of domains into regions is calculated. The greater are the regions and domains, the better is compression ratio, though the quality of restored signal may be worse.
The obtain results are discussed in the report. They indicate that the method provides rather good results in compression, though further investigations are necessary.