Squeezing A Flood Of Data Into Waves
WAVELET TRANSFORMATION IS A POWERFUL METHOD FOR compressing and analyzing information. Wavelets work their magic by representing data--including sound and pictures--as waves. These waves can be compactly described in terms of their frequency, energy, and timing. Unfortunately, wavelets remain a black art: Few people understand the complex mathematics behind them, and wavelet software is generally sold in a form that can't be readily taken apart and studied, so it's difficult to modify for optimum efficiency.
Mac A. Cody, an engineer in Richardson, Tex., is out to change that with Tsunami Plus, a library of software for performing wavelet transformations. His software is available in so-called source code in the C programming language, which means it can be easily studied or moved onto other kinds of processors. For instance, he says, an engineer could put the computation- intensive part of the program onto a dedicated, high-speed chip.
The picture shows a tutorial screen from Tsunami Plus. At the top is a sound wave. The horizontal bars below it represent the transformation of the sound by the wavelet software, with colors indicating the intensity of the "energy" in different segments of the signal. Cody says he hopes the $499 package from Mac A. Cody Associates will speed the adoption of wavelets by making it easier for nonspecialists to work with them. Wavelet transformations are already being put to use in such varied fields as storing fingerprints and recording heartbeats.