DARWIN'S CHORUS
An Interactive Evolutionary Soundscape
by Canton Becker and Jeffrey Ventrella
What's Going On Here?
Meet the swimbots. They eat, mate, and make noise. Given enough time, populations evolve the ability to locomote.
Every swimbot is born with a unique “song” and a preference for certain sonic qualities in their mates.
Use the joystick and trackball to explore these evolutionary soundscapes.
🟦 The blue buttons load up preset configurations of swimbots, demonstrating how swimbots evolve different individual and collective sounds.
🟢 The green button spawns a random swimbot. (Try using the leftmost blue button first to clear the pond.)
How Swimbots Work
The rich space-filling sound of insects and birds in diverse ecosystems inspired this sonic variation of Ventrella’s Gene Pool — an animated artificial life simulation of hundreds of swimbots (swimming robots).
In Darwin’s Chorus, each swimbot is born with a genetically-determined song and a preference for certain sonic qualities in potential mates. Over evolutionary time, musical “tribes” emerge within the pool. While the swimbots drive the evolving soundscape, we tuned the system for human listening, with sounds and tones that evoke ambient and minimalist music.
The simulation is implemented entirely in plain JavaScript with no external libraries. Dozens of concurrent MIDI channels are used to route swimbot vocalizations and background ambience to a bank of synthesizers programmed in Native Instruments Reaktor. Formant and filter modulation shape the timbre, giving the swimbot songs a vocal-like quality.
Reaktor "Swimbot Vowels" Synthesizer Patch
A Crash Course in Swimbots
- Physics + genotype–phenotype mapping. Swimbots are 2-D articulated bodies driven by oscillatory “motor” genes; separate genes set morphology (segment length/width/color/rest angles) and motor control (sine-wave phases/amplitudes/frequency), producing distinctive swimming gaits in a simplified fluid physics world.
- Selection without an explicit fitness function. There is no hand-crafted fitness score; fitness equals reproduction. Swimbots must acquire energy (green food bits) and physically reach mates. Mate preference criteria (originally, color preference, and now, song similarity and virtuosity) implement sexual selection. In this variation, swimbots are effectively blind, and can only find each other when they are making sounds.
- Emergent population dynamics. From random initial genes, poor swimmers die out while effective gaits and song structures proliferate; local “gene pools” compete for food and sex, often yielding dominant sub-populations—or total collapse—and phenomena like “breeders” vs. “sitters” under certain attractiveness regimes. We are still exploring how the introduction of sonic preferences may influence swimbot evolution.
You can play around with the original (silent) swimbot simulation at swimbots.com.
About The Artists
Since 2021, Canton Becker and Jeffrey Ventrella have collaborated on a variety of audiovisual algorithmic works. Darwin's Chorus represents their most ambitious collaboration to date.
Canton Becker is a programmer and sound artist. In addition to composing and DJing conventional electronic music, since 1995 he has built generative systems that yield unpredictable outcomes.
Becker's Pi Songs project uses the first billion digits of Ď€ to generate century-long compositions — an inquiry into musical forms that cannot be fully experienced within a single human lifespan. Later, he developed this into a whimsical stand-alone jukebox that invited participants to explore these vast soundscapes.
In Darwin’s Chorus, Becker extends Jeffrey Ventrella’s classic Gene Pool artificial-life simulation, using programming and MIDI synthesis to sonify evolutionary dynamics and make emergent structure audible in the system.
canton@gmail.com
cantonbecker.com
Jeffrey Ventrella is an algorithmic artist, artificial-life researcher, author, and educator. Since the mid-1980s he has advanced work in emergent art and fractal geometry.
In the mid-1990s he created Gene Pool, an artificial ecosystem of swimbots that evolve locomotion and mating strategies without explicit fitness functions. Widely cited in early A-Life discourse, the project offered an accessible demonstration of how complex behaviors and quasi-species can arise from simple rules and local selection pressures.
Today, Ventrella continues to produce interactive simulations and mathematical visualizations, exhibit in galleries, and publish on mathematical art and artificial life. Most recently, he published research on evolving bipedal walking figures made of spring tetrahedra.