Possibility of digital data description by means of rule dynamics in cellular automata

Nara et al. (1999) proposed a method of describing digital sound data by means of rule dynamics in two states and three neighbors cellular automats. In this paper, we report the results of evaluation of the reproducibility and the performance of describing original sounds by circulating distortion rate, power spectra and return map of signal amplitude. A typical result in our numerical evaluations gives 31.49 dB as averaged distortion rate. Comparison between the two return maps taken from the original signals and the amplitude data generated by the two states and three neighbors cellular automata shows some considerable differences, which indicates that the present method of rule extraction is not yet optimized in describing sound data. In order to improve the fidelity of describing the original data, we have tried to apply all the possible rule sequences within four rules to a certain initial bit pattern. In the best case, the trial can reproduce all the possible 16 bit binary patterns. These results suggest that the our method has strong possibility of describing digital data with high fidelity and with considerable compression.