Runs in your browser · nothing is uploaded

OCR PDF Scans

OCR a PDF to make a scanned document searchable. Each page is read with OCR on your device, and the words are added as invisible text over the scan — or copy the text out as plain text.

Language of the text

Download notice: the first run loads the OCR engine (about 4 MB) and English language data (about 3 MB) from this site. Each page is drawn at up to 300 dpi and read in a background worker on your device — expect several seconds per page.

Works best on clear, straight, printed scans. Handwriting, faint or skewed scans, tables and multi-column layouts often come out wrong or out of order — check the result.

Text found

Open a scanned PDF to read its text.

How it works

Reading the pages

The PDF is opened on your device and each page is checked for an existing text layer. Pages with text are skipped unless you ask otherwise.
Each remaining page is drawn with PDF.js at 300 dpi (lower for very large pages, so the browser doesn't run out of memory).
The open-source Tesseract engine (tesseract.js) reads the page in a background worker and returns every word with its position.
The text appears as each page finishes. Stop at any time — finished pages are kept.

What's in the searchable PDF

The original pages, untouched: the scan isn't redrawn or recompressed, so it looks and prints the same.
An invisible text layer: each word is written in Helvetica in “invisible” mode, sized to its line and stretched to the width of the word in the scan, so selection and search highlights land on the right words.
Rotated pages and pages with unusual page boxes are handled, so the text lines up with what you see.
The .txt file has every page in order: OCR'd text, plus the existing text of pages that were skipped.

Limitations, plainly

OCR makes mistakes, especially on handwriting, faint or skewed scans, small print, tables and multi-column pages. The searchable text is only as good as the recognition.
Non-Latin scripts (Chinese, Japanese, Arabic, Hindi, Urdu) come out as .txt only — the hidden text layer uses Helvetica, which can't hold them. Accented Western European letters are fine.
Sideways pages aren't straightened automatically. Rotate them in the PDF first for a good result.
Password-protected PDFs can't be read. If you know the password, remove it with Unlock PDF first.
Size: up to 200 MB and 500 pages. Long documents take a while — a few seconds per page on a laptop, more on a phone.

Frequently asked questions

Is my PDF uploaded anywhere?+

No. The PDF is opened, drawn and read entirely on your device, and the searchable copy is made there too. The OCR engine and English language data are loaded from this site. If you choose another language, its language data (a file of about 1–3 MB) is downloaded from cdn.jsdelivr.net, but only after you tick the box agreeing to it — that request fetches the language file and sends nothing from your PDF.

What does “searchable PDF” mean here?+

Your pages are kept exactly as they are — the scan isn't redrawn or recompressed — and the recognized words are added on top as invisible text, placed over the words in the scan. You can then search with Ctrl+F, select and copy text, and let other apps index it.

Why were some pages skipped?+

Pages that already have a text layer (most PDFs made from Word or a website, or scans that were OCR'd before) are searchable already, so they're left alone and their existing text goes into the .txt. Tick “Read every page” to OCR them anyway — useful when the existing text is garbled.

How do I get better OCR results?+

Start from a clean, straight scan at 300 dpi, with dark text on a light background. Choose the right language. Pages scanned sideways should be rotated in the PDF first (a PDF page organizer can do it). Handwriting, faint copies, heavy skew, tables and multi-column layouts often produce mistakes or text out of order, so check anything important.

Can it read Chinese, Japanese, Arabic, Hindi or Urdu?+

It can recognize them and give you the text as a .txt file, but it can't write them into the searchable PDF: the hidden text layer uses the standard Helvetica font, which only covers Western European letters. For English, French, German, Spanish, Italian, Portuguese and Dutch you get both.

How long does it take?+

Usually a few seconds per page on a laptop and longer on a phone, because each page is drawn at up to 300 dpi and read on your device. You can stop at any time; pages already read are kept.

Can I just copy the text out of a scanned PDF?+

Yes. The recognized text appears in a box as each page is read; press Copy to put it on the clipboard, or download it as a .txt file. You don't have to save a searchable PDF at all.

Which languages can it read?+

English, French, German, Spanish, Italian, Portuguese, Dutch, Hindi, Urdu, Arabic, Chinese (simplified) and Japanese. Choose the document's language before you start — reading with the wrong one gives poor results, especially for accents and non-Latin scripts.

Is there a page or size limit?+

Up to 200 MB and 500 pages per PDF. For a longer document, split it into parts with Merge & Split PDF, make each part searchable here, then merge the searchable parts back together.

Can I OCR a JPG or PNG instead of a PDF?+

This tool takes PDFs. For a photo or screenshot, use Image to Text (OCR), or turn your images into a PDF with JPG to PDF first and then make that searchable here.

Will OCR turn my scan into an editable document?+

No. A searchable PDF still shows the original scan; the text is hidden behind it for searching and copying, not for editing on the page. To edit, copy the text or download the .txt, paste it into a word processor, and check it against the scan.