Unicycloid and Unicycloid2 are two pieces created with variations on a one-dimensional cellular automata composition system using a cycloid synth based on the design presented in the article by Hans Mikelson "Parametric Equation Oscillators" in Csound Magazine #2 http://www.csounds.com/ezine/spring1999/
You can check out the songs at http://www.soundcloud.com/brian-wong
I probably should not have released these, there is an annoying click happening that I should have taken time to fix, but I got lazy and posted it anyway. Hopefully I will get around to identifying the culprit parameter combinations and fixing it. The microtonal serialist approach is so outside normal musical tastes I suppose I just assumed most people would hate these songs anyway, so a few clicks more or less would not much matter. :)
So far the two pieces have had 82 plays combined since being posted and no comments have been made, thus far not contradicting my aforementioned theory.
These pieces are really a rather quick first experiment with CA composition, and I hope to study and experiment with it much more. I like Unicycloid much better than Unicycloid2, but thought it might be interesting to show the results of moving one line of code which I found by accident.
There are some interesting points however to these pieces I think.
The attack time of the amplitude envelope is variable and, like all other pfields, is controlled by one-dimensional cellular automata. Note start time is adjusted according to the attack time, leading to what I think is a very useful effect.
Also adjusting note start time is a constrained-randomness using a normal distribution with a standard deviation of .0015 seconds. This adds a small amount of variation to note start times to "humanize" them.
There is no "scale" or "tuning" here. Though the initial cell is generated from a pseudo-random source, that is only for convenience. I could have used a fractal or other algorithm to fill the initial cells if I wanted a completely deterministic system.
Despite the fact that the intitial cell is random, the resulting note frequencies in the songs are not random, but related in an unusual way by the CA rules. Each note bears a definite mathematical relationship with its two neighbours, and thus after a reasonable number of generations (131 generations were used for these pieces) all the frequencies of the notes in the piece are actually somewhat mathematically related to each other. One interesting aspect of this is that I don't need to write endings or beginnings for these pieces, each song produced by the system is surprisingly "complete", in my subjective and biased opinion.
Since the mathematical relationship is localized I believe it creates a serial flow to the notes that to me is quite interesting, yet perhaps it will be undiscernible aesthetically from randomness to most listeners.