Humanities Practicum: Making Music Out of Scientific Data

A Quasar and the Night Sky Through the Eyes of L.L.A.M.A.S.
What does the universe sound like? For my final project, I wanted to communicate how the L.L.A.M.A.S. (Large Lenslet Array Magellan Spectrograph) “sees” the cosmos by translating raw astronomical data into music. I started with 21 raw, 2048×2048 spectrograph images, which inherently include a massive amount of atmospheric interference or “sky noise”. To create the background ambiance, I pixelated these images down to smaller 60×14 matrices, stacked them by wavelength (red, green, blue), and mapped the brightest pixels to MIDI notes across two octaves of a C major scale. By assigning distinct instruments to each camera, a harp for red, palm guitar for green, and a clay flute for blue, the atmospheric noise was transformed into a structured, rhythmic foundation.

The core engineering challenge, however, was isolating the target object: a high-redshift quasar. I had to manually identify the quasar’s faint spectral band, rotate it to correct a 0.75-degree tilt, and compensate for lens distortion to extract a clean 8-pixel high row of data. Because this raw target spectrum was still contaminated by the atmosphere, I isolated a nearby “blank sky” spectrum to use as a control. By carefully aligning and subtracting this control spectrum from the target, filtering out sharp atmospheric spikes, and applying a median filter, I successfully isolated the quasar’s true light signature. I then converted this refined data into a 100×14 matrix to generate the primary MIDI melody




To make the sonification emotionally engaging, I utilized Ableton Live to shape the musical dynamics. I arranged the track to build suspense, culminating in a strong crescendo when the isolated quasar melody finally entered, representing the moment of scientific discovery. A sweeping, pitch-mapped wind effect was also added to convey a sense of downward spatial movement. While writing custom Python scripts to parse FITS images and generate MIDI data was highly technical, the final result was a deeply creative exercise in finding the musicality hidden within scientific noise
