OCR PDF
Make a scanned PDF searchable and copy its text. The pages look exactly the same, and your file stays on this device.
Useful when a page is part typed text, part scanned picture. Text that is already on the page may then turn up twice when you search.
Reading takes a few seconds per page on a computer, and longer on a phone. Keep this tab open until it finishes.
Reading your document…
Everything runs on your device. Nothing is uploaded.
0%
- No uploadYour file stays on this device.
- Language data downloads onceOnly the reading data for your languages is downloaded, then kept on this device.
- Free, no sign-upNo account and no daily limit.
What happens to your file
Everything below happens inside your browser, on your own device. Your PDF or photos are never uploaded, so nobody else, including us, can see them.
- The text reader is Tesseract, a free and open-source OCR engine, running in your browser. It is loaded from pdfMB itself, not from another company’s server.
- The first time you use a language, its data (between 0.5 and 3 MB) is downloaded from pdfMB and kept by your browser. That is the only download: your file goes nowhere.
- Each page is drawn as a picture at 300 dpi and read line by line. Pages that already have real text are skipped unless you ask for them.
- The words that are found are added to the page as invisible text, placed over the same words in the picture. Search, select and copy then work in any PDF reader.
- Nothing on the pages is redrawn or compressed again. The new text is added at the end of the file, and the original file stays inside it unchanged, byte for byte.
- Photos go into the PDF as they are (JPEG photos are not compressed again), and the text is added on top of them.
- Choose a PDF or photos
- Check the languages
- Press “Read”
- Download or copy the text
How to make a scanned PDF searchable
- Choose a PDF, or one or more photos of pages.
- Check the languages. The language of this page and English are picked for you; add any other language that appears in the document.
- Press “Read”. The pages are read one by one, and you can see which page is being read.
- Download the searchable PDF, or copy the text and save it as a .txt file.
Tips for a good result
- Scan at 300 dpi if your scanner lets you choose. At lower resolutions small letters are hard to read.
- Photograph pages flat, straight on and in good light, with no shadow across the text. The Scan to PDF tool straightens pages for you.
- Turn sideways or upside-down pages the right way first, with Rotate PDF.
- Pick only the languages that are really in the document. Extra languages make reading slower and can add mistakes.
- Handwriting, very small print and decorative fonts are read poorly, so check the text before you rely on it.
When people use it
Scanned contracts and letters
Find a name, a date or an amount in a long scanned document with Ctrl+F instead of reading every page.
Old paperwork and archives
Make folders of scanned pages searchable, so your computer can find a document by the words in it.
Photos of notices and receipts
Copy the text from a photo of a notice, a menu or a receipt and paste it into a message or a translator.
Documents in more than one language
Pick two or more languages, such as Thai and English or Chinese and English, and both are read on the same page.
What it can’t do
- OCR makes mistakes, especially with blurry photos, small print, tables and handwriting. Always check names, numbers and amounts that matter.
- Text is read in horizontal lines. Vertical Chinese or Japanese text, and pages turned sideways, are not read correctly.
- The layout is not rebuilt. You get searchable text over the original page, not an editable Word document.
- Some languages read less accurately than others in our tests. They are marked “lower accuracy” in the language list.
- Language data is downloaded the first time you use a language (the size is shown before you start), so that first time needs an internet connection.
- Reading happens on your device, so it is slower on older phones, and very long documents may use up a phone’s memory. Split them into parts first.
- Password-protected PDFs, and PDFs whose owner has blocked changes, can’t be read.
Questions about OCR
What is OCR?
OCR (optical character recognition) means reading the letters in a picture and turning them into real text. A scanned PDF is only pictures of pages, so you cannot search it or copy from it until it has been through OCR.
Will my PDF look different?
No. The text is added as an invisible layer, so the pages look exactly the same. In our tests the pages are identical, pixel for pixel, before and after.
Is my document uploaded?
No. The text reader runs in your browser. The only thing downloaded is the language data, from pdfMB. Your file is never sent anywhere.
Which languages can it read?
The 32 languages of this website, including English, Thai, Chinese (simplified and traditional), Japanese, Korean, Arabic, Hebrew, Hindi, Bengali, Burmese, Russian and Vietnamese. Choose several if the document mixes languages.
How accurate is it?
On clean printed pages most languages get nearly every character right. Photos, small text and some scripts are harder. We tested every language and mark the weaker ones as “lower accuracy”.
Why are some pages skipped?
Pages that already have real text, such as pages made in Word, do not need OCR, so they are left as they are. Tick “Also read pages that already have text” to read them anyway.
Can I get just the text?
Yes. After reading, the text appears under the download button. Press “Copy text”, or “Download .txt” to save it as a plain text file.