ZUBISSOU SOUNDS · A LISTENING ROOM · BY ZUBAIR JEEWANJEE

You already know
what you want to hear.

You just don't want to go looking for it at the exact moment you're least able to — driving, mid-thought, hands full, friends arriving.

A hand-curated listening room · eight knobs · no feed

What's actually wrong

Four things, and only one of them is about music.

01The search is the work

Choosing takes the attention the moment was for. Ten minutes of scrolling to soundtrack an hour, and the hour has started without you.

02Everything sounds the same

The feed learns one version of you and plays it back forever. Familiarity arrives as comfort and stays as fatigue.

03Nobody made half of it

AI-generated tracks are arriving on the platforms unlabelled, filling the playlists that pay least and taking the pennies from people who actually play instruments.

04Every hour needs a different answer

Tuesday's focus is not Friday's dinner is not the drive home at dusk. One library, one shuffle, and no idea which part of it belongs to right now.

So: a room, not a feed.

Zubissou Sounds is a listening room on the web. Every record in it was chosen one at a time by one person. It reads the hour, the weather, the season and where your day is heading, and it opens with something that belongs to this moment.

You don't search. You turn a knob — or you don't, and the room reads the moment for you.

The part that isn't your fault

Most people stop finding new music at 31.

Not because they stopped caring. Deezer put the peak of discovery at 24 and the stall at 31; Spotify's own listening data puts the tipping point at 33. After that, the share of a life spent hearing something for the first time falls and mostly keeps falling.

24discovery peaks
31taste begins to stall
33the tipping point
24 · peak 31 · the stall 14 45 60 NEW MUSIC
Openness to new music across a life — the shape the research describes. Illustrative curve, not plotted data; the three ages on it are the measured ones.

And here is the part that matters, because it is the same sentence as the top of this page. When Deezer asked people in a musical rut why, the answer they gave most was not taste, and not age.

Underneath it there is a real mechanism. The songs from roughly 12 to 22 land on a brain flooded with hormone and dopamine while it is still deciding who it is, and they get wired in that way for good — the reminiscence bump. Nothing you hear at forty gets to arrive under those conditions. That is why the old records feel truer, and it is not evidence that they are.

So the ceiling isn't taste. It's friction — too much choice, too little time, at the exact hours you have neither to spare. Take the choosing away and the ceiling goes with it.

Sources — Deezer commissioned survey of 1,000 listeners (peak 24, stall 31, and the reasons given); Spotify listening data via The New York Times, 2018 (tipping point 33); reminiscence-bump research summarised in Slate, 2014, drawing on Daniel Levitin, Petr Janata and a 2008 University of Leeds study.

What is actually in the room

These are measured counts from what the room actually plays, as of 29 August 2026, not marketing numbers. The census is re-run daily and this page is rewritten from it.

15,847tracks
7,473artists
1907–2026years covered
95%placed on the four dials

The deepest seams are jazz (2,082 tracks), electronic (1,215), psych rock (908), asian pop (562), r&b (516), ambient (512) and a large deliberately-uncategorised “eclectic” pool (1,415) that exists because good records resist genre. The spread by decade runs 1,739 tracks from the 1970s, 1,278 from the 1980s, 1,350 from the 1990s, 1,583 from the 2000s, 2,456 from the 2010s and 4,416 from the 2020s.

About a quarter of the catalogue (3,537 tracks) has been through a BPM and musical-key census, and release years are corrected to the original release rather than the remaster date. That correction matters more than it sounds: streaming metadata routinely dates a 1969 record to its 2014 reissue, which quietly breaks any attempt to filter by era. Fixing it by hand is unglamorous and it is most of the reason the ERA knob tells the truth.

The eight knobs

Eight dials. No search box.

Every knob is a continuous axis between two named poles, not a genre checkbox. They combine, so the interesting settings are the contradictory ones — euphoric but sparse, driving but cerebral.

KnobRuns fromWhat it changes
Moodmelancholy → euphoricThe emotional temperature of the picks.
Energyambient → drivingHow hard the room pushes.
Fam · Expcanon → unknownRecords you know versus records you don't.
Erapre-1970 → todayWhere in time the room sits.
Densitysparse → wallSpace between the notes, or none.
Head · Bodycerebral → grooveWhether you are thinking or moving.
Placeglobal → specificAnywhere, or one scene in one city.
Museoff → the artistPull the room toward one artist's orbit.

There are also seven scenes that set all eight at once, for when you would rather not fiddle: Work (head-down focus), Rest (wind-down and sleep), Table (food, friends, low light), Listening (sit-down records — jazz, strings), Movement (the floor), Away (road, water, sun) and Chance (let the room choose).

How it works

Three moves, then you stop touching it.

  1. Connect Spotify

    Playback runs through your own Premium account. The room chooses; Spotify plays. Nothing is uploaded and nothing is re-hosted.

  2. Press play

    The room reads the hour and the sky and opens with three records it picked for right now. No prompt to write, no mood to pick from a list.

  3. Turn a knob when the moment turns

    Dinner becomes late-night; the drive becomes the arrival. Move one dial and the rest of the evening re-plans around it.

What the room reads

Mood is not a genre. It's a reading of the room you're actually in.

MOOD is one knob of eight, and behind it the room is reading a good deal more than where you left the dial. None of this is a setting you have to fill in. It is just true about the moment, so the room goes and looks.

The hourNot morning-noon-night. Golden hour, dusk and deep night are different rooms, and they move with the season.
The skyLive weather where you are. Rain at four on a Tuesday is not the same instruction as sun at four.
The seasonMidwinter and midsummer ask for different records at the identical hour on the identical dial.
The calendarThe holidays as they come around, and the ones near enough to lean on the evening before they arrive.
Home or outThe room learns where you usually are and notices when you aren't. Away leans a little more awake.
The day itselfA Wednesday afternoon and a Saturday afternoon are the same clock and two different appetites.
What just playedThe last few records are an argument in progress. The next one answers them, not a genre tag.
What you've heard latelyYesterday's favourite is rested, not retired. Repetition is the fastest way to ruin a good record.

What those eight signals are is not a secret — open the room and you can watch it read the sky. How they argue with each other is the whole craft, and that part stays in the box: which one wins on a rainy Sunday, when a set has gone flat and needs contradicting, how far a record may sit from the mood before it stops being the mood.

It also isn't the sort of thing that survives being explained. It is the sort of thing you put on at six on a Thursday and notice, about twenty minutes in, that you haven't touched your phone.

Take it for a spin →

How it differs from a streaming algorithm

A feed optimises for the next click. A room optimises for the hour.

A recommendation engine optimises for the next click. It learns what keeps you listening and serves more of it, which is why the fifth hour sounds like the first. This room optimises for the moment — and the shelf it draws from is finite, human and closed. Nothing gets in because it performed well. Something gets in because one person heard it and kept it.

That has a cost worth stating plainly: the room will never know a record nobody here has played. It has a benefit worth the cost: nothing in it was made by a machine, and every artist in it is a person who can be paid.

No feed. No engagement metrics. No AI music. One person's record collection, one good system, and knobs — like the Japanese listening cafés, except the collection is a 15,500-track room and it reads the sky.

Where to start with an artist

A room of more than 15,500 tracks is only useful if you can find the door into an artist you half-know. Every artist in the room has a page with the record to start on, the one to play second, and what to reach for once they've landed.

Common questions

Who made Zubissou Sounds?

Zubair Jeewanjee, one person, as a personal project. The catalogue is his; the code is his; the mistakes are his.

Do I need Spotify?

Yes, and a Premium account, because playback runs through Spotify's player. The room decides what plays; your account plays it.

Is it free?

The private beta is free. Nothing is charged for and nothing is collected beyond the address you leave to get in.

Why can't I just open it?

Spotify caps how many listeners an app in development mode may seat, so seats open in small batches. That is the only reason there is a list.

Is it like In Sheep's Clothing or Aquarium Drunkard?

Same instinct — a person with taste, pointing. The difference is that this one plays, and it reads your hour before it does.

Does it work on a phone?

Yes. It installs as a home-screen app on iOS and Android, and the knobs are built for thumbs.

Get on the list

The room is in private beta and seats open in small batches. Leave an address and the one song you love that the algorithms never play you — that second part is the good part.

Request a seat →

Your email is used for the invite and nothing else.

EVERY RECORD CHOSEN BY HAND THE ROOM · THE MOODS · THE SHELVES · THE ARTISTS · THE STORY