Mobile Music

Source: Elise Plans, “Mobile Music: A Portfolio of Works Exploring Adaptive Music Generation in Embedded and Mobile Devices” (PhD thesis, Royal Holloway, University of London, 2022). Full thesis

Claim

Composing adaptive mobile music means designing processes, mappings, constraints and software behaviour as well as sounds. The phone is part of the musical medium: its sensors, processor and operating system shape the music alongside store rules, interface design and likely obsolescence.

Evidence and method

Plans presents a practice-based PhD portfolio of generative pieces for an embedded “Breathing Stone” and mobile apps. She combines a written account of composition and software work with reflection on her own production process rather than testing a mass-market service or representative group of listeners.

She treats dataflow as a score and composition as the design and construction of processes. Manual judgement remains central because the composer writes, listens to and revises a system until its possible outputs fit a musical aim (pp. 15–16).

At BioBeats, Plans wrote endless genre-based music that reacted to cardiovascular data and became richer or more complex when a breathing exercise succeeded. The work gives one specific model of bodily data entering musical form, while also showing how app distribution, device processing and Apple approval constrain it (pp. 16–18).

Her workflow begins with timbres and samples, then moves through synthesis, weighted-random procedures and large-scale form. Plans also designs the mixer and interface before listening repeatedly for failures across the system’s possible states (pp. 27–28, 40–41).

Software forces tacit musical expectations into explicit instructions. Reusable abstractions often need revision for a specific work, and programming problems can produce both limits and useful accidents (pp. 47–50).

CPU load, app size and sensor APIs constrain the work, as do programming skill and developer collaboration. Apple review, changing frameworks and device obsolescence add further limits. Plans deliberately avoids machine learning because she wants tighter control of the possible results, which shows that generative music need not use AI (pp. 103–106).

In her responsive works, listener data can change harmony, tempo, density or available elements. Plans calls the listener’s body a co-compositional force because it produces audible changes inside a mapping the composer designed (pp. 107–108).

Concepts

• Process and dataflow as score: The composer specifies relations and behaviour through a programmed network rather than fixing every event on a timeline.
• Platform as medium: Hardware, APIs, stores, operating systems and distribution rules enter compositional decisions.
• Weighted possibility: Randomness operates within probabilities and exclusions selected by the composer.
• Long-form testing: Musicianship includes finding bad combinations, tiring repetitions and unstable transitions across many possible sessions.
• Bodily co-composition: Sensor data can alter form, though only within choices made by the system’s designers.

Limits

The thesis reports the maker’s own works and reflections, so it cannot establish how Endel’s hidden process operates or how users interpret it. Plans’ claim that the body co-composes depends on an intentional and perceptible mapping, which prevents its transfer to an opaque commercial app based only on access to health data.

Link to generative functional apps

Plans’ workflow shows the musicianship that adaptive tools require. It starts with sound families, rules and input mappings, then joins macro-form to the interface. Long listening sessions expose faults that send the composer back into the system, so generative practice changes musical skill rather than removing it.

Plans cannot establish how Endel works internally. The useful comparison is how much an Endel collaborator can shape mappings and behaviour, rather than only supplying assets for Endel’s team to adapt. Her choice to avoid machine learning also separates generative, adaptive, bioresponsive and AI-based music.