I grew up leaving cinemas the wrong way. As a kid I started music at about the age I was falling into films, and the two fused for good. I would walk out of the old Dune (the one Toto scored), out of American Pie, out of School of Rock, and the world outside the screen felt underproduced. The scenes were all still there, the corridor walk, the small victory, the crush at the counter. Only the cues were missing. Those films knew something I have never unlearned. The music was doing half the feeling, and my life had the same scenes with the sound turned off.
The adult version of me never stopped trying to fix it with consumer electronics. I am rarely without music, and I run a small logistics operation to keep it that way: the big cans for vacuuming, the sport buds for running, the little ones for the supermarket, each with its own charge level and its own pocket. And I am my own music supervisor, permanently one scene behind, because a playlist is a guess made in advance about moments that refuse to happen on schedule. The right track for the moment you are in is always one manual skip away.
I have carried the fix in my head for about fifteen years. In A Storm in a Cage I trapped a wildfire and made it brew its own rain. In The Gun That Asks First I made the police negotiate with a getaway car. This is the third round of the same game. Take the science fiction object, then climb the ladder from what exists to what does not, one honest rung at a time.
The Ghost and the Graveyard
This dream is old enough to have a ghost. Around 2010, an app called RjDj made “reactive music” on the iPhone, with the sensors and microphone feeding a score that recomposed itself around you. They even built the official Inception app, where Hans Zimmer’s music bent to your walk, your weather, your hour of the night. It was glorious, it was fifteen years too early, and it died with the company. Spotify’s 2015 Running feature belongs in the same graveyard. It matched the tempo to your stride through the phone’s sensors, and it was quietly killed in 2018. (A 2026 reboot exists. It asks you to pick a BPM in advance and does not listen to your body.)
The survivors are instructive. Video games have shipped adaptive music for three decades. LucasArts filed its iMUSE patent in 1991, and the 2016 reboot of DOOM answered a question I care about. Its celebrated metal soundtrack rises and falls in layers with the intensity of combat, so adaptive does not have to mean orchestral. Metal bends, skate punk bends, strings and horror shrieks bend. The genre your life is scored in is a catalog question, not a physics question. And Endel exists today, a real company generating ambient soundscapes from heart rate, motion and time of day.
As for 2026, demos at this year’s CES paired real-time EEG and biometric sensing with audio that responds to your state, and trend reports place adaptive audio in cars, hearables and wellness gear. As of this writing you can buy headphones that read your brainwaves, but none that adapt your music to them. The reading has shipped, but the responding has not. Everyone is building the bricks, but nobody has assembled the object. This is the moment to describe it carefully.
The Collar and the Lens
The object is an open-eared collar, with bone conduction or near-field speakers, both on shelves today. Everything else follows from the open ear, and film theory has a word for it. A movie’s score is non-diegetic. It floats above the scene without belonging to it. Earbuds seal you out of your own scene, while the collar layers a score on top of it. That is the difference between escaping your day and scoring it.
The open ear also sets the safety doctrine, and the doctrine speaks fluent mixing desk. The world is the dialogue track, and its fader is welded above yours. For sirens, alarms or a voice addressing you, the music ducks or cuts, with no override and a hard ceiling on volume. None of this is exotic. Earbuds already duck your music for conversation, European law already caps personal audio levels, and self-driving cars already listen for sirens. A fire in view cuts the score entirely.
Now the sensor problem, because the collar alone is half deaf. A heart rate tells you how much but never what. 140 BPM could be a sprint, a job interview, or a first kiss, three scenes that need three incompatible cues.
The only sensor that points at meaning is the gaze, so the second half of the instrument is a contact lens that watches what you watch. The tracking half is mature, since VR headsets steer rendering by your pupil all day. The lens half has a mixed record. The display lens died (in 2023 a famous AR lens startup shelved its lens and pivoted to selling microdisplays). The measuring lenses split. A glucose-sensing lens was abandoned in 2018, because tear glucose turned out to be a poor proxy for blood sugar, but a pressure-sensing lens for glaucoma earned actual regulatory approval. Lenses that display have ghosts, while lenses that measure have a precedent. Ours only measures.
And the lens pays twice. Pupillometry is a decades-old laboratory tool. Your pupil dilates with arousal and cognitive load, involuntarily and continuously, so the same lens that identifies the scene also reads your reaction to it. The collar hears your day, and the lens finally knows which movie you are in.
Rooms, Corridors, and the Body’s Vote
Then the music itself, and the wall nobody talks about. Your favorite songs do not bend. They are fixed recordings, and the adaptive versions a live score needs (stems, layers, intensities) are licenses the labels have never sold. That is why Endel generates its own audio instead of adapting yours.
Two engines get around the wall. The first is the pack economy, and one profession is already trained for it. Game composers have spent thirty years writing music designed to be bent by events they cannot predict. Picture an action pack, a wedding pack, a new year pack, each one a commissioned adaptive score delivered in layers, DLC for your day, in whatever genre your life runs on. The second engine composes the gaps. It is a personal generative engine trained on the shape of your library, writing the transitions and in-betweens in your own musical language so the score never jump-cuts.
Your songs are the rooms. The engine builds the corridors.
The engine never replaces the rooms. It is connective tissue, the craft DJs have always practiced by hand and streaming apps already fake with automatic blends.
That leaves the question I struggled with longest. How does the system learn which music fit which scene, when asking is forbidden? Nobody thumbs-up mid-kiss. Your body already votes, and until now the ballots went uncounted. Musical chills are physically measurable. Goosebumps are visible, skin conductance spikes at the exact second of the frisson, and a celebrated 2011 neuroimaging study caught dopamine releasing at musical peak moments. Hearts and breath entrain to tempo.
Pupil, skin, chills and entrainment add up to a coherence score, collected passively and asked directly of the nervous system. Did this music belong to this scene? The auto mode learns from every answer and fills tomorrow’s gaps a little better. This is the science fiction hinge of the essay, and I lean on it as hard as it deserves, and no harder. Every signal in the loop is real. The loop itself, worn on a neck all day, is the leap.
It also carries the essay’s darkest tension, so look at it in the light. A system that optimizes the measured chill is a dopamine slot machine, the feed-ification of your own soundtrack. The doctrine has to be one line long: optimize coherence, not chills. A score serves the scene. A drug serves itself.
The Duet
One scene, two people, should mean one cue. So the collar has a link mode. You can offer your soundtrack to the person beside you and they can offer theirs, by explicit consent, revocable with a touch. Neither one dominates, because the engines negotiate a middle in real time. The plumbing half exists. Phones already share one stream to two pairs of ears, and streaming services already compute blended tastes for two people. The negotiation half is ours.
Couples get “our song” by accident, maybe one per lifetime, minted by a radio coincidence. The collar mints them live. The cue that played while you both watched the same storm come in now belongs to two nervous systems. And the moment either of you revokes, the duet dissolves politely back into two solitudes. Nobody’s day gets annexed.
The Right to Silence
The loop has a thermostat problem, and I will not hide it. The sensors read your state, the music changes your state, the new state changes the music. The collar is a thermostat, not a mirror, and somebody sets the target temperature. It had better be you, in settings you actually understand, and not a mood curve optimized by whoever sold the pack.
There is a precedent for the accusation this device will face. Within a few years of its 1979 launch, the Walkman triggered a moral panic. The charge was that it would privatize public space, with everyone walking around scoring their own life. The collar is the Walkman taken all the way, and the old charge deserves a straight answer. Yes, that is what it is, without the sealed ears, with the world welded on top, and with consent at every layer.
Last comes the feature I now believe matters most. Ask film composers about their most powerful cue and many will give the same answer: the cut to silence. A score that never stops is wallpaper. So the collar’s finest skill is knowing when to shut up. The same sensors can detect the scenes too large to touch, and the collar steps back so the moment stays entirely yours. The machine’s real mastery lies less in picking the perfect track than in recognizing the moments that deserve none.
Here is one fragment you could start this afternoon, in the spirit of this site. Your phone can already tell running from vacuuming from a walk to the store. A dumb DJ app that switches between three of your own playlists on that signal, and ducks hard for conversation, is a weekend of work. It would answer the first real question. Does a scored day feel like a film, or like a cage? If you build that, tell me.
So here is what I keep turning over. If every scene of your life arrived with the right cue, would you still trust what the scene made you feel on its own? And which moments would you fence off, unscored, so they stay entirely yours?
