Two Thirds of a Mix Stays Unidentified: Data From 904 Real SoundCloud Mixes

Across 904 real SoundCloud mixes, two thirds of playtime stays unidentified even after audio fingerprinting and comment analysis. The data behind track ID culture.

There's a comment sitting under a mix right now. It says: "ID at 47:12? That bassline is killing me." It was posted on a Tuesday. It's Saturday. Three people have liked it. Nobody has answered it.

You know this comment because you've written it. You've hovered over the timestamp, replayed the same eight bars until the transition blurred, typed the question, and then waited. Checked back the next morning. Checked back at lunch. Told yourself you'd stop checking, and then checked again from bed.

That wait — the itch of an unresolved track ID — is one of the strangest and most beautiful rituals in music culture. And it deserves to be talked about with respect, because it's easy to mistake it for a problem. It isn't. It's a community doing what databases can't.

We measured it: two thirds of a mix stays unidentified

That itch is not a niche experience, and now we can put a number on it. We took every SoundCloud mix that real listeners have run through TrackRadar's analyzer — 904 unique mixes, 1,450 hours of audio — and measured how much of each mix came back with an identified track, and how much stayed unknown.

The analyzer splits a mix into roughly two-minute segments and runs each one through three audio fingerprint services, then reads the description, the comments, and the tracklists people have already posted. Across 42,272 segments, this is what came back:

  • 67.9% of all segments produced no identified track — pooled across every mix in the corpus.
  • The median mix had 65% of its playtime unidentified (mean: 64%).
  • In 75% of mixes, the majority of the playtime stayed unknown.
  • In 8.2% of mixes, at least 90% of the playtime came back with nothing.
  • The middle half of mixes sits between 51% and 79% unidentified — this is the normal range, not the extreme.

In other words: on a typical mix, machines plus the answers humans have already posted resolve about a third of the playtime. The other two thirds is exactly the space where the "ID?" comment lives. If you are new to the notation itself — what "ID - ID" in a tracklist means — start with our guide to what "track ID" means.

Method and population

Population: 904 unique SoundCloud mix URLs analyzed by real listeners through TrackRadar, measured September 1, 2026 — one analysis per unique mix URL (the most recent), with demo content and internal test accounts excluded. A segment counts as resolved when one of three fingerprint services or the text analysis (captions, comments, posted tracklists) returned a track for it. Two caveats worth stating plainly: a long track spanning two segments is counted in one, which slightly overstates the unidentified share, while overlapping blends pull the other way. And "unidentified" means unidentified by our system — a human with the vinyl may know exactly what it is. That gap is the point of this article.

How the ritual works

The mechanics are simple and almost universal across platforms. Someone uploads a mix, a radio show rip, or a thirty-second clip of a set. Someone else hears a record they can't place. They ask. Usually just those two letters: "ID?"

Then the trading begins. A stranger replies with a partial clue: "Sounds like something on a Dutch label, mid-nineties." Someone else adds a fragment of the vocal, half-remembered. A third person swears they own it on vinyl but the sleeve is at their parents' house. Days pass. Sometimes weeks. Then — if the thread is lucky — someone drops the full answer: artist, title, label, year, maybe a pressing note. The original asker replies with more gratitude than most people show for a birthday gift.

This isn't new. Electronic music forums have run dedicated ID threads for decades — Dogs On Acid, a drum & bass community founded in 2000, still hosts running "Track ID" threads that span dozens of volumes, each one a slow-built ledger of questions that took years to answer. The venues have moved from forums to blog comments to today's comment sections under mixes and clips. The ritual hasn't.

Why it persists

Here's the thing outsiders miss: the slowness isn't a flaw. It's part of the value.

Track ID culture runs on reciprocity. The person who answers your question today asked their own question three years ago and remembers what it felt like to be answered. Reputation accrues to the people who know — the ones whose name in a thread means the search is over. Being that person feels good in a way that's hard to explain to someone who has never dug for records.

And there's a harder, practical truth underneath the social one: obscure records live in human memory before they live in a database. A white label pressed in a run of three hundred copies, a track that only ever existed on a promo CD-R, an unreleased edit a producer played twice — for records like these, a collector's memory is often the only index that exists. No piece of software replaces that. It shouldn't try.

So the culture persists because it works, because it feels good, and because for the deepest cuts it's genuinely the best tool available.

But it has a cost, and you've paid it: the unanswered question. The thread that goes quiet. The mix you bookmarked in 2023 with a mental note that still reads "find that track at 47:12."

Where a recognition step fits

This is where we think an audio-recognition step belongs in the workflow — not instead of the thread, but before it and alongside it.

The move is simple. You take the link to the mix or the post — TrackRadar works with links from platforms like SoundCloud, YouTube, Mixcloud and Instagram, or with an audio file you upload — and paste it in. The audio gets fingerprinted, and the surrounding text (captions, comments, tracklists people have already posted) gets read too, because sometimes the answer is sitting in plain sight three hundred comments down. What comes back is a set of candidate tracks, each linked out to places like Spotify, Bandcamp, Apple Music and Discogs so you can verify with your own ears.

Candidates, not verdicts. Some records won't match — the white labels, the unreleased edits, the deepest of the deep. That's not a failure of the ritual; that's where the ritual takes over. The thread still wins those.

But here's what changes: instead of arriving at the comment section empty-handed, you arrive with something to trade. "Ran this through recognition — 47:12 comes back as this, can someone with the vinyl confirm?" You've turned a question into a contribution. You've given the thread a starting point instead of a blank. And when the tool does land the answer outright, you get to be the person who closes someone else's five-day wait in five minutes — which, let's be honest, is the best feeling this culture has to offer.

The hunt stays social. It just gets a head start.

An invitation, not a pitch

If there's a mix that's been sitting in your bookmarks with an unresolved timestamp, try this: paste the link into TrackRadar, listen through the candidates, and take whatever you confirm back to the comment section where the question lives. Answer someone else's "ID?" this week while you're at it.

The culture built the index one thread at a time. The least a tool can do is help you feed it faster.

TrackRadar