The Song and the Fingerprint: What I Listen for in the Age of AI Music

EN SV
The Song and the Fingerprint – AbeArtsSweden

A song can now be made in the time it takes to brew coffee. Type a mood, a genre and a few lines about heartbreak or summer, and out comes something with a verse, a chorus, a bridge, and a voice that sounds like it has lived a little. The first time you hear it, it is startling. The second time, something quieter happens: you start listening for what is missing.

I have been trying to name that missing thing. Not to win an argument about whether machine-made music is “real” — that debate is loud and mostly circular — but because the answer says something about why I make music, and why I love it at all.

My working answer is this: every song that stays with me is keeping a secret. And a good listener is someone who notices it.

The fingerprint

The music I return to always has a fingerprint. A place where the singer’s voice almost cracks. A chord that is slightly “wrong” and therefore right. A lyric so specific it cannot have been written for everyone — only for someone.

A fingerprint is not perfection. It is evidence. It shows that a person was there, deciding, hesitating, choosing this and not that. Detectives look for fingerprints because they prove presence. Listeners do the same thing, mostly without knowing it.

AI is superb at the smooth surface of a genre. It knows what a Nordic folk ballad, a late-night synth track or a stadium anthem is supposed to sound like, because it has learned from thousands of them. But a model learns by averaging, and the average of everything is, by definition, no one in particular. What I listen for now is the deviation: the moment a piece stops being typical and starts being somebody’s.

The secret inside the song

Here is a simple way to picture it. Imagine two letters. One is a perfectly written template that says all the right things. The other is shorter, a little clumsy, and mentions the blue cup you always drink from. The first letter is correct. The second one knows you.

Songs work the same way. The secret is not hidden to be clever. It is the private detail the writer could not leave out: a street name, a melody borrowed from a grandmother’s kitchen, a rhythm that came from the sound of a train. You may never learn what it refers to — but you can feel that it refers to something. That feeling is trust. It is the listener sensing that a real life sits underneath the sound.

A machine can imitate the shape of a secret. It can write “the blue cup.” What it cannot do is have one, because there is no kitchen, no grandmother, no train behind the words. The detail is decoration, not memory. Sometimes the difference is invisible. Often, over time, it is not.

Friction is part of the meaning

There is a reason we love the story behind a record: the tape hiss from a borrowed room, the take that was only kept because the neighbour started mowing the lawn in rhythm. Limits and accidents are not obstacles to meaning. They are where a lot of it comes from.

In plain terms, friction is anything that slows the maker down and forces a choice. A studio with only one microphone. A singer who cannot reach the high note and has to find another way. One evening left before the band splits. Each of these pushes the song somewhere it would not have gone on its own.

Remove the friction, and we get speed and abundance — and often a strange weightlessness: music that sounds finished but feels like it could have been anything. That does not make the tools useless. It makes the question sharper: where do I want friction, and where am I happy to skip it?

What radio taught me about listening

Before Sweden, I spent years in radio — though rarely in front of the microphone. Most of my work happened behind the programmes: as a recording editor, as a correspondent sending in reports, and as a facilitator, making room for other people’s voices to come through.

Editing teaches you to listen differently. You sit with the raw recording and decide what stays. Which pause to keep. Which breath to cut. Which stumble to leave in, because it is the most honest moment of the whole interview. A clean edit that removes every hesitation often removes the person, too. The best edits kept a little of the roughness, because that was where the listener could hear someone real.

As a correspondent and facilitator, I learned the other half. Listeners forgive a lot — a small mistake, a pause, some noise on the line. What they do not forgive is the feeling that nobody is really there. A voice reading a script and a voice actually speaking to you can use exactly the same words, and the audience always seems to know which is which.

That lesson has come back to me in the age of AI music. The question was never whether the sound is polished. The question is whether anyone is home — and whether the one doing the choosing left the fingerprint in.

Where I let the machine in

I make music with AI tools myself, so this is not a sermon from outside the room. For me the line is not “AI or no AI.” It is whose choices end up on the record.

A sketchbook, a sparring partner, a way to hear an idea in five arrangements before committing to one — that is a tool in the hands of a maker. Handing over the choosing itself is different. If I cannot say why a line is in the song, it is not really my song yet.

  • Use it to explore: rough textures, alternate keys, the idea you would otherwise abandon.
  • Keep for yourself: the feeling that started it, the judgement of whether a line is true, the final yes or no.
  • Be honest about it: listeners can forgive almost anything except feeling misled about who was in the room.

Where my songs begin

People sometimes imagine that making music with AI means typing a sentence and pressing a button. For me it begins much earlier — and it begins with a kind of pain.

I choose a particular issue in society, sometimes local, sometimes global, and I stay with it. I read about it and watch it with focus, until it is no longer just news to me but something I carry. I discuss it, too, including long conversations with AI, turning the issue over and asking what it means for the people living inside it. Slowly, the facts turn into feeling. That is the part no model can do for me: absorbing the weight of someone else’s situation until it becomes empathy.

Only at the culminating point, when the feeling is full, do I ask the machine to help. I ask it to extract the substance of everything we have gone through together and transform it into lyrics and melody. The machine is fast and skilled at form. But what it shapes is something I brought into the room.

So every song in my collaboration with AI carries this kind of pain inside it. You may not know which news story, which image or which conversation started it. That is the secret. But if you listen closely, I hope you can hear that someone was hurting on behalf of others before a single note was generated.

This is the triangle I work in: the human, the machine, and the work between them. I bring the empathy and the final judgement; the machine brings form and speed. The work is only honest when both sides of that triangle are visible — and when the feeling at its centre is real.

And there is one place where I keep the machine out entirely. My memoir series, Testimony, is written without AI, on purpose. Craft can be refined with a tool. Witness cannot. Some secrets have to stay in the hand that lived them.

A question for the listener, too

As the feed fills with flawless, frictionless sound, listening becomes a small act of choice as well. Streaming platforms reward what keeps us playing, and a model can be optimised for exactly that. There is nothing wrong with background music — there never was. But it is worth noticing when we are being soothed rather than met.

Do I want something tuned to be inoffensive, or something made by a person who risked being misunderstood? Both have a place. I just want to know which one I am pressing play on.

The test

So this is my current test. Not “was a machine involved?” but “can I find the fingerprint?”

When I can, I lean in. I listen again, trying to hear the secret — knowing I may never learn what it is, only that it is there. When I cannot, I move on, gently, and without needing the song to be a villain.

In a world where sound is endless and cheap, presence becomes the rare thing. The fingerprint is how a song tells you: someone was here.

What do you listen for? Tell me about the last song that made you feel someone was really there — and whether you ever found its secret.


← Back to Music