replay console·11 recorded runs·latest capture 2026-06-12 PDT

Find the feeling in the track.Not the crowd around it.

Give it a song you love and it finds others that genuinely sound like it. It starts from the songs listeners tend to play together, then listens to each one and reorders them by how they actually sound.

pick a song·11 saved runs to try

play

Each song here is a real run you can play back, step by step. Running a fresh one makes a real Anthropic API call, roughly a cent or two each, so these are saved playbacks rather than live runs to keep the costs down. The deep dive has the details.

the crowda saved run · Midnight City by M83 · 0:13the sound

how it runsquick preview · real time · write-up not shown

The crowd0:01
180 to consider · 75 we could hear
The sound0:00
listened to 75
The final list0:00
top 10 picked · +0:12 write-up

what it picked · top 5

1Steve McQueen by M830.883
2Trying to Be Cool by Phoenix0.820
3Tongue Tied by Grouplove0.808
4Safe and Sound by Capital Cities0.803
5Kids by MGMT0.785

picks saved 2026-05-29 PDT (work in progress), timings recorded 2026-06-12 PDT

watch the whole run, step by step →

Live · the real engine

connecting…

The real engine, right now. This part is live, unlike the saved runs everywhere else on the site.

API

—

Songs known

—

Searches run

—

Last backup

—

Independent monitor (healthchecks.io): healthchecks.io status

Getting the latest numbers the server sent…

The short version

Finding songs that genuinely sound alike is harder than it looks. Here’s the idea.

01
The problem
“Play me something like this” usually just means “play me something popular.” Most apps go by what other people clicked, not by what the song actually sounds like.
02
Why the easy fixes fall short
You can’t just ask a chatbot what a song sounds like, because it has never heard it. And a fixed library of free-to-use tracks answers a hit song with thirty tunes nobody knows.
03
What Doppel does
It uses both sides. The crowd is good at knowing which songs get played together. is good at knowing which ones sound alike. Lean on each for what it does well, and the picks get a lot better.
04
Why you can trust it
None of this is hand-waving. Every result page lets you watch the list narrow down with the real numbers behind it, and see how much the listening step reshuffles the crowd’s picks.