PDF OCR

A scanned PDF is a picture of a page — it contains no text, which is why you cannot select or search it. OCR recognises the shapes as characters and adds a real text layer behind the image, so the document looks identical but becomes searchable, selectable and convertible.

Open it and try to select a line of text. If words highlight, it already has a text layer and OCR is unnecessary. If your cursor draws a rectangle and selects nothing, it is a scan and needs OCR first.

English, Hindi (Devanagari), Bengali, Tamil, Spanish, French, German and several others. Choose the correct language before running it — accuracy drops sharply when the engine is looking for the wrong alphabet.

Very high on clean, straight, reasonably high-resolution scans of ordinary printed type. Accuracy falls on faint photocopies, unusual fonts, tight tables and pages photographed at an angle. Always proofread numbers — a misread digit looks entirely plausible.

Partially, and unreliably. OCR is trained principally on printed type. Neat block capitals are sometimes recognised; ordinary cursive generally is not. Treat any handwritten result as a draft to check rather than an extraction to trust.

No. Recognition runs inside your browser using WebAssembly. This matters here more than almost anywhere else on the site, because the documents people scan are contracts, certificates and identity papers.

You get a searchable PDF that looks exactly like the original. From there you can convert it to Word or Excel, copy text out of it, or simply keep it as an archive that can actually be searched.