OCR - make a PDF searchable

No. The scanned image stays byte for byte as it was; what is added is an invisible text layer above it, the same construction scanner software uses for a searchable PDF. Search, select and copy work; the print is unchanged.

It is Tesseract 5's neural engine, the same one behind most open-source OCR, running as WebAssembly in your browser. A clean 300 dpi scan of printed text comes out almost letter-perfect; a skewed, faint or photographed page less so - straighten it first with the Straighten tool. Words the engine wasn't sure about are still placed, so a search still finds them, and the tool says how many there were.

Hungarian, English, German, French, Italian, Spanish, Polish, Romanian, Slovak, Czech, Croatian, Slovenian, Serbian (Latin and Cyrillic), Ukrainian, Russian, Bulgarian, Greek, Turkish, Danish, Swedish, Norwegian, Finnish, Dutch, Portuguese, Indonesian, Vietnamese, Lithuanian, Latvian, Estonian, Malay, Latin, Arabic, Persian, Hebrew, Chinese (simplified and traditional), Japanese, Korean and German Fraktur for pre-war blackletter print, in any combination - a bilingual contract can be read with two. Each language is a separate download the browser then keeps.

Nothing is uploaded: the file is opened and processed inside your browser and never leaves your device. Close the tab and it is gone. The OCR engine and its language files come from this site too, so not even the choice of language is sent anywhere.

By default only those with no text layer yet, because a born-digital page already has one and a second would double every search hit. If an earlier OCR did a poor job, switch to every page - the new layer goes under the old one.

Longer than anything else here - recognition is real work, and every page takes a moment even on a quick machine, so a long scan is worth starting and leaving to run. The first run also fetches the recognition engine and the language data (a few megabytes, from this site rather than a CDN); after that your browser has them cached, and the button shows how far it has got, page by page.

Yes. Drop in a .docx, .xlsx or .pptx file or an image (JPG, PNG, WebP...) and it is turned into a PDF in your browser as it arrives - an image onto an A4 page - then handled like any other PDF; nothing is uploaded. The conversion is our own, so the result can differ slightly from Office's own PDF export: a font may be substituted and charts are left out. The file list shows a warning when something couldn't be carried over.