The debate about AI-generated music versus human-made music usually circles around art. But there is a quieter, more practical side to it: functional music, the kind you put on to support concentration, rest or sleep. Here, the question is not whether a machine can write a hit. It is whether generated audio can do the specific, detailed work this kind of music demands.
This article looks at what changes when music is composed and played by people, what research says about music built for attention, and why one upcoming app – Sotuno, a functional-music app from Cologne launching in 2027 – works exclusively with human artists.
AI-generated music vs human-made music: what is actually different
Generative music systems are trained on large catalogues of existing recordings. From those patterns, they produce new audio that sounds plausible: chord progressions resolve, drums stay on the grid, textures fill the frequency spectrum. For casual background listening, that is often enough.
Functional music has a harder job. It is built to shape your state over 25, 50 or 90 minutes – to help you settle into a writing session, stay with a long analysis, or let a racing mind slow down before sleep. That requires decisions across long timescales: how a piece opens, when energy rises, how a transition is prepared so it never startles you, when a motif returns so the music feels coherent rather than random. If you want a fuller picture of this category, the guide on what functional music is and what it isn't covers the basics.
Current generative systems are strongest at local plausibility – the next few seconds sound right. They are weakest at exactly what functional music depends on: deliberate structure over long stretches, with a purpose behind every change.
What gets lost when no one is playing
Micro-timing and touch
When a person plays an instrument, no two notes are identical. A pianist leans slightly ahead of the beat to create momentum, or sits behind it to relax the feel. A cellist shapes each bow stroke. These variations are tiny – milliseconds, small shifts in dynamics – but your ear registers them as warmth and intention. Generated audio can imitate variation statistically, yet it varies without meaning. Over an hour of listening, many listeners find that the result feels flat or vaguely tiring, even if they cannot say why.
Structure with a purpose
A composer writing for concentration makes choices you never consciously notice. Keep the melodic movement predictable so it never pulls focus. Avoid lyrics, because language competes with the same mental resources you need for reading and writing. Place a gentle lift where attention typically dips. Each choice answers one question: what should the listener's mind be doing right now? A generator has no such question. It has only the statistics of its training data.
Accountability
Human-made music has an author. If a track works, the artist can build on it; if a transition is too abrupt, it can be revised. Generated catalogues are produced in bulk, and no one is responsible for any single track. For music you rely on daily, that difference is not sentimental – it is a quality mechanism.
What research says about music built for attention
The scientific picture on background music is nuanced, and it is worth knowing before you trust any app, human-made or generated. A meta-analysis by Vasilev et al. (2018) found that background speech and lyrics tend to interfere with reading and memory tasks – one reason serious functional music avoids vocals entirely. Woods et al. (2024, Communications Biology) reported that music with rapid amplitude modulation was associated with better sustained-attention performance than unmodulated control audio, suggesting that specific acoustic properties, not just "music in general", do the work.
Two things follow. First, the details matter: modulation, tempo, texture and the absence of lyrics are design decisions, not accidents. Second, none of this is a blanket promise. Effects differ between people and tasks, which the overview on whether music helps you focus explains in more depth. The honest position is that carefully designed music may support concentration for many listeners – and that careful design is easier to guarantee when identifiable people with scientific input are making the choices.
For listeners with ADHD traits, the details matter even more. Research on ADHD and music suggests that stimulation level and predictability play a large role in whether sound helps or distracts. That is the kind of parameter a composer can tune deliberately – and a bulk generator cannot. If ADHD affects your daily life, sound is at most one part of a routine; it is not a substitute for professional diagnosis and care.
How Sotuno works with people instead of generators
Sotuno, developed by Kiwimo-Product GmbH in Cologne, has made this a founding rule: every track in the app is composed, played and mixed by established artists who bring science and music together. Software still has a role – it arranges, extends and mixes the recorded material so a session can run seamlessly for as long as you need – but it never generates a note. The distinction is deliberate. Arrangement is logistics; composition is judgement.
The app offers four modes – Focus, Relax, Sleep and Meditate – plus a single switch called ADHD mode, which changes how every session is built: more drive, faster modulation, shorter blocks with a clear start, no sudden changes, and an intensity dial to adjust the level of stimulation. Those parameters echo what attention research points to, and they only work if the underlying music was written with them in mind. There are no lyrics, no ads and no playlists to manage.
Transparency runs through the rest of the product as well: your data stays on servers in Germany in ISO 27001-certified data centres, with no ad networks and no selling of data. Sotuno is also preparing its own medical studies with clinical partners for 2027. It has not published studies of its own yet – and says so plainly.
Questions to ask any music app before you rely on it
Whether you end up with Sotuno or something else, a short checklist helps you judge what you are actually listening to:
- Who made the music? If an app cannot name its composers or artists, assume the catalogue is generated or licensed in bulk.
- Is there a design rationale? Look for concrete acoustic choices – no lyrics, controlled modulation, prepared transitions – rather than vague talk of "scientifically optimised sound".
- Are the science claims specific? Real evidence names studies. A claim like "proven to boost focus by 40 percent" without a source is a warning sign. The same scepticism applies to popular audio formats; the piece on binaural beats shows how far marketing can drift from evidence.
- What happens to your data? Free background-music apps often finance themselves through ads and tracking. Check where your data is stored and whether it is shared.
A provider that answers all four questions openly is rare. When you find one, the music tends to reflect the same care.
AI-generated music versus human-made music is not an abstract culture debate when you listen for hours every day. Functional music lives on long-range structure, purposeful detail and accountability – all things that come from people making decisions, not from statistical plausibility. That is why Sotuno works exclusively with artists who combine musical craft with scientific grounding, and why it names its evidence instead of inventing it. If that approach matches how you want to work, rest and sleep, the early-access list is open now, and joining is free.
