The Expression of Evolution Through Sound (ETS) system is an exploration of the process of Genetic Algorithm (GA) based composition. Unlike conventional Genetic Algorithm (GA) composition tools whose goals are to use genetic algorithms to create aesthetic music or sounds, the Evolution Through Sound (ETS) system is an attempt to express the actions of a GA in an aesthetic manner. This audification is done in the hope that its structure forms an interesting sound experience. Individual frequencies of an ETS population are evolved for harmonic relationships with each other, and the entire population of solutions is played back simultaneously through the combination of sine waves. The result is a complex wave that evolves from white noise into an interesting audibly harmonic waveform that continues to change with time.

This web page serves as a guide to the ETS system. On this page you will find:

• A description of the ETS concept.
• Audio samples of the output by an example ETS system.
• Source code in C (major) for the ETS jMax object.
• A jMax example of a system containing the ETS object.

Downloads

The ETS System

The ETS system was developed in the jMax music programming environment. The core of ETS is a genetic algorithm object capable of interfacing with other jMax objects. The object outputs several lists of frequencies, each corresponding to one segment of the population (we'll get into this below). These frequency lists are then de-multiplexed and sent to sine wave generator objects. The modularity of ETS makes it potentially useful in other genetic algorithm systems. The output of the ETS system is the complex wave created from the combination of the sine wave generators controlled by the population, along with a number indicating the total score of the population at each generation.

Settings

• l : The lifetime (number of generations) an individual member of the population lives for. Increasing the lifetime of an individual makes the system more stable
• p : The size (number of individuals) in each generation. A maximum population of 25 individuals (per generation) is recommended.

Other system settings can only be modified through the recompilation of the source code. I'd like to add some of these to the object instantation, and probably will in a later version. Settings that are modifiable include:

• Mutation Rate: the probability that the mutation operator will be applied to a newly created member of the population.
• Fractional Ratios: the desired ratio of fractions between the frequencies.
• Ratio Scores: the corresponding score of each ratio.
• Margins: the margin of deviation from a perfect fractional relationship allowed. Narrow margins make the harmonic relationships hard to find, but the sound created is truly harmonic. Wide margins produce cyclic patterns known musically as beatings that, although not as harmonic are very interesting.

Current Work

The ETS system has recently undergone user testing. The suggestions of the user testing are currently being examined and some are being implemented. Some of these suggestions include features that would allow the system to:

• Focus in on specific frequencies or octaves using a pass-band filter.
• Control the number of sine wave generators that are output by the system dynamically.
• Filter out the noise of the system by controlling the percentage of the population (based on score) that gets audified.

Links

The following links were useful in the development of the ETS system:

• A good survey paper of GA Composition.
• A jumping off point for GA Composition examples on the web.
• A good tutorial on consonance and dissonance.
• An interesting consonnance application.
• An overview of temperment and modern musical concepts.
• A description of the Mesopotamian temperment used in the ETS system.

Contact Information