Generative systems in music therapy

March 30, 2026

Written by Jordan Elias, MT-BC

What happens when a composer (or therapist) steps back from decision making and creates a process instead? Two tools above let you explore this directly: one inspired by Alvin Lucier's iterative room recordings, one by Éliane Radigue's slowly evolving drones. The article below traces the ideas behind both, and considers what generative music has to offer as both an artistic philosophy and a therapeutic framework.

What is a generative system in music?

Most music in the Western tradition involves a composer specifying what sounds should occur: which notes, in which order, at which volume, for how long. The performer's job is to realise those specifications as faithfully as possible.

With generative music the composer creates a set of rules, conditions, or processes, and then allows those conditions to unfold without further intervention. The music emerges from the architecture of the system. Brian Eno, who coined the term, described the experience of setting up generative systems and then encountering the output as a listener rather than an author, often surprised by what their own rules had produced.

This is not the same as improvisation, where a performer makes real-time decisions. It is not the same as chance music either, where a composer uses dice or random operations to determine what happens. Generative music is deterministic but unpredictable because the interactions between elements produce results that were not explicitly composed. The composer designs the conditions for music to happen.

I want to share two composers who explored this: Alvin Lucier, who made the physical environment itself a generative instrument, and Éliane Radigue, who composed through the slow evolution of electronic drone. As a therapist I think therapy can be thought of in a similar way. Therapists don't always produce change directly; we aim to create the conditions in which change becomes possible. Through looking at these composers I want to reflect on how music therapy can adopt a generative music mindset and explore some ways to incorporate this music into the therapeutic process.

Room as instrument

In 1969, Alvin Lucier sat in a room and recorded himself reading a short text describing what he was about to do. He then played that recording back into the room through a speaker, and re-recorded the sound in the room. He played that recording back, re-recorded it again, and repeated the process dozens of times. In each iteration, the room acted as a filter that reinforced certain frequencies corresponding to its physical dimensions and absorbing others. Gradually, his voice dissolved. The room's resonant frequencies, determined by the size of the space, became audible as pure sustained tones.

What makes the work so interesting to me is that the room is a compositional instrument. Every enclosed space has resonant frequencies called room modes: standing waves that form when a sound wave reflects back and forth between surfaces and arrives back in phase with itself. The lowest mode depends on the longest dimension of the room: in a room 5 metres long, the lowest axial mode is around 34 Hz. These modes are normally inaudible, but Lucier's iterative process amplifies them. With each iteration the frequencies of the room become more prominent, and others disappear.

The text Lucier reads during the recording begins: "I am sitting in a room different from the one you are in now." The words transform over iterations into sustained resonances that eventually lose their semantic content.

Listen to I am sitting in a room.

Try it yourself

The Lucier module above simulates this process using digital filters. But the original process is even more interesting to try with the natural resonances of real rooms.

Record yourself saying a sentence or short phrase. Play it back through your phone speaker in a room. Record that playback on another device, like a laptop microphone. You now have the first iteration. Play that recording back in the same room and record it again on the first device. Repeat this four or five times. If you are consistent with keeping the speaker and the recording microphone in roughly the same positions, you will hear the room's resonant frequencies beginning to emerge. The closer the recording microphone is to a wall or corner, the more the room modes will be reinforced, because those are the locations where standing waves have maximum amplitude.

A few variables to experiment with: different rooms (bathroom, stairwell, and large hallway), different source materials (speech, singing, a single sustained note), and different distances between speaker and microphone. Each configuration produces a different outcome because it engages the room's acoustic properties differently.

The art of slow time

Éliane Radigue composed a lot of music with an ARP 2500 synthesizer and later with acoustic instruments, where the sound unfolded over extremely long durations and with changes so gradual that individual moments of transformation are difficult to locate.

Radigue came to music through Pierre Schaeffer and Pierre Henry, two composers foundational to musique concrète. She worked in Henry's Paris studio as an assistant before developing her own practice. She was also deeply influenced by Tibetan Buddhism and the connection between her musical philosophy and Buddhist ideas about time, impermanence, and sustained attention is explicit in her own statements about the work. She described composing as a form of meditation practice, attending to the smallest fluctuations in the sound with the same quality of presence she brought to formal practice.

The technical basis of her drone compositions is simple: two or three oscillators tuned to nearly identical frequencies, whose slight differences produce slow beating, or an amplitude modulation at the difference frequency. If two oscillators are tuned to 110 Hz and 112.5 Hz, they produce a beat at 2.5 Hz. The effect is a slow, barely perceptible pulse. Try it yourself here. As the oscillator frequencies drift slightly over time, the beat frequency changes, and with it the entire texture of the sound. The analogue noise floor of her vintage equipment added a thin layer of texture that gave the tones a sense of materiality and presence.

Listen to L'Île re-sonante.

Generative systems and therapy

In Acceptance and Commitment Therapy (ACT), a core process is called cognitive defusion, or the practice of creating distance between a person and their thoughts so that thoughts can be observed as thoughts rather than taken as literal truth. When someone is strongly fused with a thought, such as "I am inadequate," it functions not as a belief that could be questioned but as a transparent fact about reality. Defusion techniques create perceptual distance: the thought is repeated until its meaning fades, or observed as passing weather rather than permanent climate, or heard as words rather than experienced as statements about the world.

The Lucier process is a physical enactment of defusion. A sentence is run through a process that gradually strips away its meaning until only its resonant structure remains. What was language becomes sound. The piece does not argue with the sentence's meaning, does not challenge or reframe it. It simply runs it through a system of attention and repetition until the meaning falls away and something underneath becomes audible. An exercise I find interesting is using this iterative audio processing with a client's own recorded voice as a way of creating literal distance from the content of negative self-statements, running the Lucier process on words that carry painful associations and listening as those words become frequencies. This is not a technique I would use with everyone, or early in a therapeutic relationship. But for some people, at the right moment, the perceptual experience of meaning dissolving is more compelling than any cognitive reframe.

The central practice of Mindfulness-Based Stress Reduction is to simply direct your attention to what is happening now, with curiosity and without judgment. When attention wanders, notice that it has wandered and return. The skill being developed is not the absence of mind-wandering but the noticing of it, and the capacity to return without self-criticism.

Radigue's music demands exactly this quality of attention. Nothing announces itself. There are no melodies or rhythms to follow. What there is: a sustained field of sound that changes so slowly the changes are perceptible only if you are genuinely present for consecutive moments. If your attention wanders and returns, you will find yourself unable to locate the change that occurred in your absence. The music rewards the same thing mindfulness practice develops: non-grasping, present-moment attention without expectation of dramatic content. And it requires a different relationship with time, one where something can be present and valid without progressing toward a conclusion.

A neurodiversity-affirming approach

Many traditional therapeutic interventions were developed with neurotypical nervous systems in mind. They assume certain things about how people process sensory information, regulate emotions, respond to repetition, and experience distress. For many neurodivergent individuals, particularly those who are autistic, have ADHD, or experience sensory processing difficulties, those assumptions do not hold.

Exposure therapy, for example, is built on the principle that repeated contact with a feared stimulus will lead to habituation; the nervous system will learn that the stimulus is not actually dangerous, and the fear response will diminish. This works well for anxiety disorders rooted in inaccurate threat perception, but for many neurodivergent people, distress does not arise from inaccurate beliefs about danger. It arises from genuine sensory or cognitive overload. The texture of certain fabrics or the flicker of fluorescent lights can cause real distress because their nervous system processes sensory input with less filtering and more intensity than a neurotypical system does.

Repeated exposure to an overwhelming sensory stimulus without control does not produce habituation. It produces increased dysregulation and sometimes trauma. Really the goals should be regulation and agency. A neurodiversity-affirming approach asks whether an environment has the right conditions for a person to thrive. Generative systems operate under a similar philosophy; instead of correcting the output, we redesign the system.

Experiments

A note on therapeutic applications

Designing good conditions and then releasing control of the outcome is both an artistic practice and a relational one. Generative systems unfold at the pace of their internal dynamics. Therapy happens within a trusting relationship between therapist and client. In both cases, progress is hard to see in the moment, but unfolds over time.

Working with generative systems requires a tolerance for uncertainty. Once the process begins, the composer relinquishes control. Therapy demands a similar attitude. Insight can't be scripted, nor can a therapist guarantee an outcome. But they can aim to maintain conditions under which change becomes more likely.

The tools on this page are available as a standalone application with additional features on GitHub.

If this resonates

This post is part of a series on sound, perception, and experimental music as therapeutic tools. Earlier posts cover how overtones shape timbre and what EEG research actually shows about binaural beats. I also have an earlier post about indeterminate music.

If you are curious about what music therapy grounded in these ideas looks like in practice, feel free to reach out via the contact page. I work with musicians, neurodivergent individuals, and people navigating significant life transitions. No music experience required.

I offer a free 30-minute consultation to explore whether working together might be a fit.

Further reading:

• Deep Listening — Pauline Oliveros
• Acceptance and Commitment Therapy — Hayes, Strosahl, Wilson
• Full Catastrophe Living — Jon Kabat-Zinn