Seeing Music

Music has never been content to stay in the ear. It borrows a body the moment it can find one. A foot on a wooden floor. A hand on a table. Someone's shoulders in a crowd, moving a half-second before yours do. Long before anyone built a spectrum analyser, we were already turning sound into something to look at — because hearing alone was never quite enough for it.

The rooms remember this better than we do. Candles behind coloured glass in a temple. A mirror ball, which is only a lamp and a lot of small mistakes. Winamp, on a beige computer, at two in the morning. Every one of these is the same instinct wearing different clothes: give the sound a place to be seen.

There is even a small, forgotten art form for it. In the 1920s a man named Thomas Wilfred built an instrument that played nothing but light, and toured it as seriously as a pianist tours a concerto.^[1]1. Thomas Wilfred called his instrument the Clavilux and his medium Lumia. He performed silent light recitals in concert halls in the 1920s. Nobody remembers, which is its own kind of proof that this impulse keeps having to be reinvented. No sound at all. Just slow colour, folding over itself, in front of an audience that had bought tickets. It never caught on, and it never entirely went away either. It keeps getting rebuilt, in every decade, by someone who has just noticed that their music has nowhere to look.

§The bar chart problem

Somewhere along the way, the visualizer became a readout. Bars. Columns. A waveform crawling left to right like a hospital monitor. These are not pictures of music — they are pictures of measurement. They tell you, faithfully and boringly, how much energy sits at 4 kHz. And because they are honest instruments, they are entirely deaf to the thing that matters: how the song feels while it is happening to you.

Anything that moves in perfect lockstep with the beat stops being interesting after about eight bars, because you can predict it. Prediction is the death of looking. What holds attention is something that has its own life and is merely disturbed by the music — a weather system with a song blowing through it. Ink that keeps drifting during the quiet part. A flock that was going somewhere before the bass arrived and will still be going somewhere after.

That was the whole design rule, and I set it before I wrote a line of shader code: never a bar chart. The sound is not the subject. The sound is the wind.^[2]2. The technical version of this argument, and the eight scenes that survived it, are written up on the project page. This one is the reason underneath.

§Why I built it

I built Dhun — धुन, melody — because I wanted my desktop to know what I was listening to. Not as a notification. Not as a widget with a progress bar and a rounded rectangle. I wanted the album art to just be there, a small square of colour floating over my work, changing temperature when the song changed, the way a room changes when someone opens a curtain.

The honest reason is smaller than any of that: I have spent most of my working life inside a screen full of text, and I wanted one thing on it that was not asking me for anything. A record that spins because a record should spin. A wallpaper that is tonight's album instead of a mountain I have never visited. Ornament, deliberately — the thing software spent twenty years apologising for and quietly removing.

There is a specific pleasure in software that is useless in the economic sense and indispensable in the daily one. Dhun does not make me faster. It cannot be sold — I gave up that right on purpose, in exchange for being allowed to adapt other people's beautiful shaders. It has no growth curve, no roadmap that ends in a company. It is 1.4 megabytes of decoration, and I open it every single day.

§Where it goes

Nowhere in particular, and that is the plan. This is a hobby project in the oldest sense: built for an audience of one, extended whenever the one gets curious. Half the visualizers I wrote never shipped — they were beautiful for a week and then I stopped looking at them, and not-looking is the only review that has ever mattered to me. What survived is what I still notice on a Tuesday.

So I will keep going. More scenes, when a good one arrives. Better hands on the controls — an onset detector that hears a snare and not just a kick, something that knows the difference between a build and a drop. Maybe, eventually, the visualizer becomes the point and the player becomes the accessory. I have no idea. That is roughly the entire appeal.

Sound gets to move through a room and disappear. Everything else on my screen is permanent, indexed, and waiting for input. It seems fair to let one small square of it be temporary and beautiful and about nothing.