Science has many languages: mathematics, words, equations, diagrams, images. Each has its own geometry, and shapes what we think through it. Prose unfolds as a sequence in time; a map spreads information across space. Music can hold many temporal processes at once, letting them coexist, interact and transform without reducing them to a single line. Sounds of the Sciences began with a simple question: what happens to scientific ideas when we think them through music?

The project is a collaboration with neuroscientist Dr. Mariano Sigman. So far it comprises six pieces of music, each with its own visualization, portraying a scientific idea, a field of science, or a moment of discovery. Some are built from data; others emerge from running simulations.

The aim was never to take something already understood and translate it into sound. We wanted to find out whether music might reveal aspects of an idea that are harder to grasp in words, numbers or images. Not a more accurate understanding, and not a replacement for an equation or a diagram, but a different one: something closer to a felt sense of the idea.

Early on we ran into a distinction. Were we making music about nature, or about science? Science is a human activity with a history. The planets did not change when their orbits came to be described mathematically; what changed was our understanding of them. That history of understanding runs through several of these pieces, and it is why the plural in the title matters to us.

The way we composed also came to resemble the scientific process itself, with its repeated testing and reformulation of hypotheses. Each piece began with a what-if question: What if the motion of the planets generated rhythm? What if evolution took place among melodies and rhythms rather than organisms? An idea became an algorithm, the algorithm made music, and then we listened to the result. Sometimes it failed, sometimes it succeeded; either way, it suggested the next hypothesis.

The pieces are neither scientific demonstrations with musical decoration nor compositions loosely inspired by science. Their scientific structures constrain the music without fully determining it, and where the data leaves a choice open, musical thinking steps in, often revealing unexpected aspects of those structures. We hope they offer a visceral appreciation for science and the scientific process.

All six were made in Python, using SCAMP, my open-source library for computer-assisted composition. Many thanks to the James S. McDonnell Foundation for supporting this work.

The piano states a simple integer sequence based on binary counting (OEIS A319419). Around it grow layers of rhythm, each filtering out a different hidden property of the numbers: divisibility, twin primes, self-similarity. A single line of numbers becomes a polyphonic texture.

This is also a piece about a problem. The pattern is catchy, but it is boundless, doubling in length and growing in width with each added digit, with no natural place to stop. Much of the composing became a series of experiments against its repetitiveness. One was an orchestration governed by powers of three, set at cross-purposes to the sequence's powers of two so that changes fall across its seams. Another gave each twin prime a long sustained note that intensifies every time. In many ways, the music is a record of our attempts to constrain this unruly beast and bring it, ultimately, to a close.

We can conceive of infinity, but we can never manifest it. Infinity tries anyway, using the Cantor set. The set is usually described by division: take a line, remove its middle third, then the middle third of each piece that remains, forever. The music builds it the other way around, by addition: a note, then a copy of it after a rest, then a copy of that whole pattern after a longer rest, and so on. The result is a self-similar rhythm of notes and rests.

The difficulty is that the Cantor set is almost entirely empty. Each duplication adds more silence than sound, and the rhythm grows ever sparser. To counteract this, the tempo accelerates constantly, in what is known as a Risset rhythm: the rhythmic counterpart of the Shepard tone, whose pitch seems to rise forever. Several copies of the pattern sound at once, each nested inside the gaps of the one before. Heard from a distance, the separate layers merge into a single broad moment, the way the notes of a strummed guitar chord become one gesture.

What we actually perceive, in the end, is a finite, human slice of an infinite pattern, much as we see only a narrow band of the electromagnetic spectrum as visible light. The set itself says nothing about pitch, so we superimposed an eight-note melodic pattern on the sequence of notes. This turned out to make an enormous difference, allowing the listener to sense where they are within each superimposed layer of the sequence.

The visualization shows both constructions side by side: the additive one the music is built from, and the divisive one, the line split into thirds again and again. The same form can be read as infinite combination or as infinite division.

A musical simulation of the present and (somewhat fantastical) future of our solar system. As its observers, we keep time by Earth's orbit, each year a measure and each beat a season, the way humanity kept time by the sky for millennia. Because the orbit is slightly elliptical, even this pulse runs a touch faster at some points than others. Into it the other planets gradually enter, creating a throbbing polyrhythmic dance. The outer planets pulse basslines in a simplified ratio of their orbital periods, where Jupiter and Saturn's 2:5 relationship is clearly audible. Later, with the greater resolution afforded by mathematical astronomy, we begin to hear the planets' elliptical melodies and the gravity that drives them.

The solar system is many simultaneous motions at very different scales, hard to hold together in a diagram but natural for music to sustain at once. A literal simulation of its distant future, though, would outlast any piece. So the model is gradually pushed: the Sun falters, and the planets' pull on one another is strengthened to compress long timescales of disruption. Where the scientific model ends and the musical fiction begins is hard to say, and that ambiguity is part of the point. Eventually the orbits lose their integrity and collapse, and the steady pulse of Earth and its inhabitants drifts off into the void.

Implemented in Python, using solar system data from astropy.

A musical interpretation of Conway's Game of Life, a grid of cells governed by a handful of simple rules. As the system evolves, different beings emerge from the evolutionary soup (blocks, beehives, blinkers, gliders), each with its own musical expression. It's almost miraculous that something so "alive" can come out of so little.

That threshold is the heart of the piece. Rather than sonifying the raw grid, the program listens for the moment a pattern becomes recognizable as a life form. Entities are detected using convolution and tracked from frame to frame. Each kind has its own attack, sustain and release gestures, panned by its horizontal position and transposed by its vertical one.

The montage seen in the video superimposes several outputs of the program, assembled and spatialized in Reaper. Programmed in Python using pygame and SCAMP, which communicates with SuperCollider for the audio rendering.

What if evolution took place among instruments rather than organisms? Several species of musical creature live in an environment of two measures of 12/8 time. Food, shown in green, is distributed to reinforce the standard accent patterns of 12/8. The creatures begin silent; as they start to play, those that find food are deemed fitter and reproduce more, and the beat comes into focus.

The piece turns on exaptation, the repurposing of a trait for something new. Feathers, for instance, appear to have served dinosaurs as insulation long before they were adapted for flight. Here, genes that coded for rhythm begin to find expression as melodic and harmonic instruments. Then a predator arrives. Over many generations, creatures that feed on the beats where it is active are weeded out, and as the predator grows more active, the playing grows sparser. When it suddenly disappears, it leaves an evolutionary ghost: sparser, more tentative rhythms, traces of a selective pressure that is gone.

In one last act of exaptation, the genes that once coded for beats are reinterpreted as pitch classes. Evolutionary pressure now acts on harmony and pushes toward consonance, and the population settles into a looping pentatonic chord progression. It seems to come out of the blue, but it has been coming all along.

Dmitri Mendeleev is credited with discovering the periodic table in 1869, but the classification and ordering of the elements was a collaboration across many centuries, by many scientists. Nor did the elements become periodic when Mendeleev arranged them. What changed was our ability to perceive an order in observations that were still incomplete. This movement tries to reproduce something of that experience.

We sample the periodic table at several moments in history, each pass set a little later, watching the initially sparse table fill out as new elements are discovered. Each time, we hear it through the lens of a new property, taken roughly in the order science learned to measure them: specific heat, atomic radius, electronegativity, radioactivity. The properties take turns as soloists while the others accompany, and at the end they all sound at once. At first the music can seem sparse, almost arrhythmic. As more elements arrive, periodicities begin to emerge: a pattern that eventually sounds obvious, but was at first hidden by the gaps in our collective knowledge.