Skip to content
PDF Carpenter

Make a scanned PDF searchable

A scanned PDF is really a stack of pictures of pages, so you cannot search it, select a sentence or copy a paragraph out of it. OCR, short for optical character recognition, reads the text in those pictures and adds it back as an invisible layer behind each page image. The pages look exactly the same, but you can now search for a word, highlight a passage and paste it elsewhere. Pick the language of the document before you start: English, German, French, Portuguese and Simplified Chinese are available. Recognition is slow work, so documents are limited to 30 pages. Clear, straight, well-lit scans give the most accurate text.

Processed on our server, deleted after 10 minutes

Where is my file processed?

This tool needs a server. Your file is uploaded (up to 95 MB), processed on our infrastructure and deleted 10 minutes after the job is created, or as soon as you start over.

How it works

  1. 1

    Add your scanned PDF

    Drop the PDF onto the page, or choose it from your device. It must be within the page limit shown on this page.

  2. 2

    Choose the language

    Select the language the document is written in. A wrong choice is the most common cause of poor results.

  3. 3

    Set the options

    Decide whether pages that already contain text should be left alone, and whether slightly tilted pages should be straightened first.

  4. 4

    Start OCR and download

    Press the button and wait, as recognition can take a few minutes. Then download the searchable PDF and try searching for a word.

Frequently asked questions

Which languages are supported?

English, German, French, Portuguese and Chinese (Simplified). Choose the one that matches the text on the page. Handwriting is not recognised reliably in any language, and neither is text in a language that is not on the list.

Is there a page limit?

Yes. A document can have at most 30 pages and be up to 95 MB. For a longer document, split it with Split PDF, run each part through OCR, and merge the results afterwards.

Why is there a limit on how often I can use it?

OCR uses far more computing power than most tools, so each visitor can start only 3 OCR jobs per hour, and jobs may wait in a queue when others are running. The page shows your place in the queue.

What happens to pages that already have text?

By default pages that already contain real text are left as they are, and only the image-only pages are recognised. Choose the option to process every page if the existing text is wrong or unusable. The look of the pages does not change either way.

How accurate is the result?

It depends on the scan. Sharp, straight, high-contrast pages in a supported language come out well. Faint print, skewed pages, unusual fonts, stamps and handwriting cause mistakes. Search the result for a few words you know are on the page to judge it before relying on it.