THE BODY CONDUCTS

Translating Chronic Stress into Music

Chaeyeon Kim · Sep 2025 – Apr 2026

Composition · EEG signal processing · Python · Spatial Audio

TL;DR

The Body Conducts is a five-track music film that translates EEG data to music through a Python-based biometric-to-audio pipeline. I recorded my own EEG under controlled stress, processed the raw signal in Python, and drove the score’s sound directly from that data. The invisible physiology of chronic stress becomes audible.

Drawing on Prokofiev’s Peter and the Wolf, each instrument represents a distinct stress response. The score spans orchestral adagio, dialogic oboe-synth arrangements, a capella, and a full noir jazz brass breakdown, with the composer’s own brainwave (EEG) data embedded directly as rhythm and texture. The result makes invisible inflammation audible and inescapable.

Project playlist Code

The Body Conducts uses spatial audio. Recommended with headphones.

THE PROBLEM

Think about this scenario:

you have been quietly enduring workplace stress,

so your baseline heart palpitations go up,

so your immune system reads this as physical danger

and floods your body with inflammation,

as if preparing for a wound that never comes.

Do this daily for months, and your body starts attacking itself.

Resources on stress might tell you to meditate more, exercise, and manage your time better. But when stress comes from social hierarchies or structural inequality, these individual behavior changes are a band-aid on a structural wound. It risks scientifically misrepresenting where chronic stress originates, and lets the systems causing harm off the hook.

The biological mechanism

By the time you feel pain, invisible inflammation has been building for years.

Philosopher Elaine Scarry argues that chronic pain has no referential content, which means you hurt but cannot point to why. Language fails the body in pain… but emotionally resonant narratives don’t. Music, uniquely, triggers physiological chills and makes you perceive the experience.

In short, to communicate a threat no one can see, you make them hear it.

THE SOLUTION

I built a full biometric-to-audio pipeline and used it to produce a finished five-track music film.

Biometric Integration Pipeline

  1. 01

    Data collection

    • Wore a 4-channel OpenBCI EEG headset (200 Hz) during stress-inducing scenarios (N-back tasks, social-evaluative stress) to capture neural activity.
    • Alpha waves (8–12 Hz) during baseline, Beta waves (12–30 Hz) during stress.
    • 40+ hours of recording across sessions.
    EEG data collection session with OpenBCI headset, electrode paste, and live signal monitoring
  2. 02

    Signal processing & feature extraction

    • Using Python (MNE, SciPy, NumPy), preprocessed the raw signals and extracted time-series features: frequency-band power and amplitude variation.
    • Applied 4th-order Butterworth bandpass filtering and sliding-window log band-power extraction on the alpha band, then derived a continuous neural index from band-power dynamics relative to a baseline condition.
    • Also engineered STFT spectrogram analysis across recording conditions to visualize spectral shifts over time.
  3. 03-A

    Spectral filter mapping

    • Mapped data to a parameter that correlates to sound.
    • The stress signal (a continuous 0–1 value derived from alpha suppression) was exported as MIDI CC data using the Mido library and imported into Logic Pro as an automation lane controlling a dynamic spectral EQ filter.
    • As stress rises, the filter sweeps toward brighter frequencies, increased dissonance, and denser texture; as it falls, the sound becomes warmer, more consonant, and more open.
  4. 03-B

    Direct wavelength translation

    • Mapped data directly to notes.
    • I transposed the beta-wave patterns directly into the jazz piano lick in Rot, creating a transcription of brain state into musical phrase.

Spatial Audio Rendering

The completed score was mastered in Dolby Atmos using Logic Pro’s 3D audio engine, placing individual sound sources independently across left, right, above, behind, and below the listener. For a project about chronic social stress, which is atmospheric, ambient, and arrives from no single locatable source, this was a deliberate decision: the spatial grammar of the audio format itself encodes the phenomenology of the stress the film depicts.

Dolby Atmos renderer: binaural objects placed around a listener in a 3D room
Dolby Atmos · binaural renderer

THE SIGNAL

The pipeline reads a continuous stress index (derived from alpha suppression) that drives the spectral filter, alongside the spectral shift between a calm baseline and an active stressor (alpha power drops, beta power rises).

stress-index curve · alpha suppression vs baseline
band-power comparison · relative 4–40 Hz
Circular STFT heatmap of recovery EEG, shown as a record
STFT heatmap · recovery, as a record

THE TRACKLIST

Act Track Genre / instrumentation Character EEG condition
I Overture Distorted orchestral Wolf (all, merging) distorted Wolf theme
II A440 Oboe ↔ synth-pop Duck & Cat dialogic recording
III Prima A capella Bird (pre-trauma) baseline · pre-stress
IV Rot Noir jazz brass Wolf (stress) composite stress avg.
V Finale Adagio → jazz Full orchestra + Wolf average of all data

Each track is a movement in the conductor’s arc, and each carries EEG data from a specific recording condition.

THE RESULT

WHAT THIS PROJECT DEMONSTRATES