DJs are visual people. Scroll through any DJ's Instagram feed and alongside the club videos and set announcements, you will find something distinctly analog: photos of vinyl records. A freshly arrived shipment from Juno. A rare find at a second-hand shop. A stack of 12-inches laid out on a turntable, labels facing up, waiting to be played.
For other music fans, these posts are tantalizing. You can see the record. You can sometimes make out the artist name or the label. But actually identifying the tracks on that LP from a photo? That used to require squinting at tiny text, manually searching each title, and hoping you read the label correctly.
TrackRadar now does this automatically. Point it at a photo of a vinyl record label, and it identifies the tracks -- artist, title, and links to Spotify, Bandcamp, Apple Music, and Discogs. This is the story of how we built it.
The Problem: Two Very Different Label Layouts
When we first started working on vinyl detection, we assumed it would be a single problem: read text from a circular label. In practice, we quickly discovered that vinyl labels come in two fundamentally different layouts, and each requires a different approach.
Layout A: The numbered tracklist. This is what you see on most LPs and compilation records. The label has a clearly printed tracklist with side markers and track numbers: A1, A2, B1, B2, and so on. The artist name usually appears at the top, and the tracks are listed below in order. This layout is relatively structured and predictable.
Layout B: The minimal single. This is the format used by most 12-inch singles and EPs in dance music. There are no track numbers. Instead, you get the track title in large text and the artist name below it -- or sometimes the other way around. The label might also include the record label name, a catalogue number, and not much else. These layouts are far less consistent and sometimes ambiguous about which text is the artist and which is the title.
Any system that only handles one layout would fail on half the records DJs actually post. We needed to handle both.
AI Vision That Reads Like a Record Collector
Our approach uses AI-powered image analysis trained specifically on vinyl record labels. When you submit a photo, the system first determines which layout type it is dealing with. For numbered tracklists, it extracts the structured data: side, number, and title. For minimal singles, it uses contextual clues -- text size, position, and label conventions -- to distinguish between artist names, track titles, and metadata like catalogue numbers or mastering credits.
The system handles real-world photography challenges that you might not immediately think about. Vinyl labels are circular, so text often curves along the edge. Photos are taken at angles, under varying lighting, and sometimes with reflections from the glossy surface. The AI accounts for all of this, reading curved text and compensating for perspective distortion.
We also built in validation to catch common misreadings. If the system extracts an artist name that does not exist in any music database, it flags the result as uncertain rather than confidently returning wrong data. This matters more than you might think -- a single misread character in an artist name can send you down a completely wrong search path.
From Label to Library: The Linking Pipeline
Reading the label is only half the job. Once we have extracted artist and track names, we need to find those tracks across streaming and purchasing platforms. This is where TrackRadar's multi-platform search comes in.
Each extracted track gets searched simultaneously across Spotify, Bandcamp, Apple Music, Discogs, and SoundCloud. Our smart matching system handles the inevitable discrepancies between how a track appears on a vinyl label and how it is listed on a digital platform. Record labels often use different formatting conventions -- a vinyl label might say "Feat." while Spotify uses "ft." or "(with)" -- and our matching accounts for these variations.
Discogs linking is particularly valuable for vinyl identification. When someone posts a photo of a record, there is a good chance they want to buy or sell it. The Discogs link takes you directly to the marketplace listing for that specific release, not just the track.
How It Holds Up on Real-World Photos
In our early spot-checks we ran vinyl detection against real Instagram photos of vinyl records, covering everything from pristine studio shots to blurry phone photos taken in dimly lit record shops. In the large majority of those checks the text on the label was extracted correctly -- we have not yet run a controlled benchmark large enough to put a precise number on it.
The photos that fail are almost always extreme cases: severely out-of-focus shots, labels where the text is almost entirely obscured by wear or stickers, or rare pressings with hand-stamped labels that are barely legible even to the human eye.
For comparison, manually identifying tracks from a vinyl photo typically involves zooming in, guessing at partially visible text, searching each guess individually, and cross-referencing with Discogs. For an LP with 8 tracks, this can easily take 10 to 15 minutes. TrackRadar does it in under 10 seconds.
Built for Crate Diggers
Vinyl detection was built specifically with crate diggers in mind. If you spend your weekends flipping through stacks at record fairs, this feature changes the game. See a record you do not recognize? Snap a photo, paste it into TrackRadar, and instantly know whether those tracks are worth your money.
It is also powerful for Instagram discovery. When a DJ you follow posts their latest vinyl haul, you no longer need to ask "what's that record?" in the comments. TrackRadar identifies the tracks from the photo and gives you links to listen immediately.
You can try it now with the Vinyl Record ID tool -- just upload a photo or paste an Instagram URL containing a vinyl record image.
If you are a collector or crate digger, we have also put together a dedicated guide for getting the most out of TrackRadar's vinyl features at our crate diggers use case page .
What Comes Next
Vinyl detection is live now and included in all TrackRadar plans, including free accounts. We are already working on improvements: better handling of colored and picture disc labels, support for cassette tape J-cards, and the ability to scan multiple records in a single photo.
Vinyl might be analog, but discovering what is on those records does not have to be. Take a photo, let TrackRadar read the label, and start listening.