OCR PDF — Scanned PDF to Text

OCR PDF reads the words off scanned pages and returns them as a text file, with a confidence score for every page.

Each selected page is rendered to a canvas at about 144 dpi and handed to Tesseract, compiled to run inside the browser, which recognises characters from the pixels and reports how certain it was; any page scoring under 70 per cent is named individually so you know which ones to check against the original. It reads one language at a time, English unless you say otherwise. The limit to understand before relying on it: the output is a .txt file, and the PDF you gave it comes back unchanged with no invisible text layer added — this makes the words available to you, it does not make the document searchable.

Recognition runs on your own device, the engine included, so neither the page images nor the recognised text is sent anywhere for processing.

Getting the figures off a scanned invoice without retyping them, finding a clause in a photographed contract, lifting text from a scanned book page, and judging whether a scan is legible enough to be worth keeping.

Also known as: searchable pdf · scan to text pdf · recognize text in pdf · ocr scanned pdf · image to text · photo to text