If you collect techno, trance, or house, you know that vinyl labels are a beautiful, chaotic mess.
Between the giant label logos, the tiny copyright warnings, the obscure remixer credits, and the complete lack of standardised formatting, building a scanner to read physical records is a nightmare.
Standard barcode scanners fail on white labels, and traditional OCR (Optical Character Recognition) just spits out a wall of unreadable gibberish. This was a big problem in the development of crAte!
After repeatable attempts to implement the Barcode, CAT# scan with a fallback OCR scanner for the inner circle label, I realised that a basic OCR scanner with Google Vision API integration was not enough. I successfully captured ALL the text from the lable but it was really bad at actually identifying records. I needed a smarter way to scan records and digitise a collection.
So, after a lot of brainstorming, ideation, and coding mishaps (with a bit of help from Gemini) I built a “Vinyl Intelligence” engine for the crAte app.
Instead of relying on brittle pattern-matching, crAte now uses a two-step pipeline. First, it uses Google Cloud Vision to extract every single piece of typography off the spinning wax. Then, it passes that raw, messy text to a lightweight AI parser (Gemini 2.5 Flash) with one strict job: act like a professional archivist.
The AI instantly filters out the RPM speeds, the rights society acronyms, and the distributor noise. It surgically extracts the primary Artist and Title, and drops them directly into your collection.
It turns the tedious process of manual data entry into a single tap. The crAte app is getting smarter, and digitising your physical collection has never been faster. Stay tuned for the upcoming beta launch.
