ERRORLESS DESCRIPTION WITH TWO RULES OF CELLULAR AUTOMATA FOR DIGITAL SOUND DATA

Abstract

We investigate errorless and perfectly reproducible coding of real sound data by means of one-dimensional cellular automata with two states and three neighbors related with the viewpoint of "rule dynamics". In our previous studies, we have shown that only two rules in 1-2-3 CA are sufficient to generate all the patterns consisting of 16 bits. Moreover, it was shown that the digital data of several spoken words and music can be reproduced perfectly. However, the reason why two-rules set of 1-2-3 CA can generate errorless coding for digital sound data has been unknown. In addition, the potentiality of the description method, data compressibility and description ability, has been unknown. In the present paper, therefore, we present the proof that the two-rules set can generate arbitrary bit-pattern sequences. From computer experiments, the two-rules set of (#90, #180) is practical in realizing compressive coding for digital spoken-words and music data without reproducing errors.