Cellular automata (CA), like every other dynamical system, can be used
to generate music. In fact, starting from any initial state and applying
to them simple transition rules, such models are able to produce numerical
sequences that can be successively associated to typically musical physical
parameters. This approach is interesting because, maintaining fixed the
set of rules and varying the initial data, many different, though correlated,
numerical sequences can be originated (this recalls the genotype-phenotype
dualism). Later on a musification (rendering) process can tie one or more
physical parameters typical of music to various mathematical functions:
as soon as the generative algorithm produces a numerical sequence this
process modifies the physical parameter thus composing a sequence of sounds
whose characteristic varies during the course of time. Many so obtained
musical sequences can be selected by a genetic algorithm (CA) that promotes
their evolution and refinement.
The aim of this paper is to illustrate a series of musical pieces generated
by CA.
In the first part attention is focused on the effects coming from the
application of various rendering processes to one dimensional multi state
CA; typical behaviours of automata belonging to each of the four families
discovered by Wolfram have been studied: CA evolving to a uniform state,
CA evolving to a steady cycle, chaotic and complex CA. In order to make
this part of the study Musical Dreams, a system for the simulation and
musical rendering of one dimensional CA, has been used.
In the second phase various CA obtained both by random generation and
deriving from those studied in the first part are organised into families
and, successively, made evolve through a genetic algorithm. This phase
has been accomplished by using Harmony Seeker, a system for the generation
of evolutionary music based on GA.
The obtained results vary depending on the rendering systems used but,
in general, automata belonging to the first family seem more indicated
for the production of rhythmical patterns, while elements belonging to
the second and fourth family seem to produce better harmonic patterns.
Chaotic systems have been seen to produce good results only in presence
of simple initial states. Experiments made in the second part have produced
good harmonic results starting mainly from CA belonging to the second family.