Read text from images and scans
Screenshots, photos of paper, scanned PDFs — pulled into selectable, searchable text by OCR that runs on your device.
How it works
- Drop an image or scanned PDF
PNG, JPG or a PDF — scanned pages are rendered and read page by page.
- Let it read
First use downloads the recognition engine (~11 MB, once). Then it works through the pages, with progress shown.
- Take the text
Copy it, download as .txt, and fix the inevitable couple of misreads — 'rn' versus 'm' remains a human specialty.
Questions people actually ask
Which languages?
English for now. More languages mean more megabytes, so they're added one at a time rather than all at once.
How accurate is it?
Clean digital renders and sharp scans: excellent, better than 99% on normal text. Phone photos of curved paper: entirely readable, with a few corrections needed.
Is the text going anywhere?
No. Recognition runs in the tab, on your machine — which is also why it takes real seconds instead of a server's instant score. The trade is understood and accepted here.