OCR PDF
Read the text in scanned pages and add a text layer, so you can search, select and copy it.
Select a PDF fileor drop it anywhere on this page, or paste
Opens in your browser. Nothing is uploaded.
OCR PDF, free and private
OCR PDF turns a scanned PDF into a searchable PDF. A scan looks like text but is a picture of a page, so you can't search it or copy a reference number. OCR (optical character recognition) reads the letters in that picture and adds them as an invisible text layer behind the page image.
Recognition runs in your browser with Tesseract. The first time you use a language, the engine and that language's model are downloaded. Your PDF is never uploaded, and there is no page cap.
How to make a scanned PDF searchable
- 1
Add your scanned PDF
Press Select PDF files or drop the scan on the page. The tool handles one PDF at a time.
- 2
Pick the document language
The panel shows which pages already have text and which will be read. Under Language of the text, search for the language and pick it, or up to three if the pages mix languages. Each model downloads the first time you use it.
- 3
Run OCR
Press Run OCR. Each page without text is drawn at 300 DPI and read in turn. Pages that already have selectable text are skipped. Expect a few seconds per page, depending on your device.
- 4
Download and check
Press Download PDF to save the file ending in -ocr.pdf. Search for a few words you can see on the page to confirm they were picked up.
Why use PDFToolz for this
Recognised on your device
Tesseract runs in a Web Worker in your browser. The only downloads are the engine and language files. Your PDF's content is never sent anywhere.
Read at 300 DPI
Each scanned page is drawn at 300 DPI before recognition, the resolution Tesseract is tuned for. Small print and thin letters read better than at lower resolutions.
Digital pages left alone
Pages that already have selectable text are copied as they are, with their original quality and text. In a file that mixes scans and digital pages, only the scans are read. If every page has text, the panel says there is nothing to OCR as soon as you add the file.
Over 90 language models
From English, Spanish and Chinese to Arabic, Hindi and Greek, and up to three at once for mixed pages. The right model improves accuracy because it knows that language's letters and common words.
How to OCR a PDF accurately
OCR accuracy depends mostly on the scan. Clean, straight, evenly lit pages of printed text read well. Faint photocopies, skewed photos and busy backgrounds give more errors. A few minutes of prep helps more than anything else.
- Straighten crooked scans with Deskew Scanned PDF, and turn sideways pages upright with Rotate PDF, before you run OCR.
- Pick the language of the text. Each run uses one language, so a file mixing English and German reads best in whichever language there is more of.
- When scanning paper with a phone, fill the frame with the page. Scan to PDF can crop, straighten and clean up phone photos first.
- Handwriting, decorative fonts, tiny print and complex tables are the hardest cases and may come out incomplete or jumbled.
What changes in the searchable PDF
The tool rebuilds the document page by page. Each scanned page is redrawn as a 300 DPI image with Tesseract's text layer over it, at the original page size. Pages that already had text are copied unchanged. Links, form fields, bookmarks and document properties are not carried over. The file is often bigger than the scan.
To shrink the result, use Compress PDF on Light or Recommended. Both keep the new text layer. Avoid Extreme, which turns every page into a plain image and drops the text.
If you can already select text on every page, you don't need OCR. Use PDF to Text or PDF to Word to pull out the text that is there, which is faster and exact.
Proofread OCR text
OCR is a best guess. Common slips: similar shapes (l, 1 and I; O and 0; rn and m), missing accents, merged columns and misread numbers in tables. Search only finds what was recognised, so a misread name will not turn up.
Before you rely on copied amounts, dates, account numbers or legal text, compare them with the page image. In long documents, spot-check pages from different parts of the file.
Frequently asked questions
Add the PDF, pick the language and press Run OCR. You get a copy with a hidden text layer you can search and copy from. It is free and runs in your browser.
No. Your PDF is drawn and read in your browser. The only network requests fetch the OCR engine and language model, which hold nothing about your document.
Over 90 languages, including Latin, Cyrillic, Greek, Arabic, Hebrew, Indic, Chinese, Japanese and Korean scripts. Pick up to three when a document mixes languages.
They are skipped and copied as they are. Only pages with no selectable text are read. If every page already has text, the tool says there is nothing to OCR and suggests PDF to Text.
Very close. Scanned pages are redrawn as 300 DPI images, and pages with text are not touched. Links, form fields and bookmarks are not kept.
Not reliably. Tesseract is built for printed text, so handwritten notes usually come out incomplete or wrong.
The first time you use a language, your browser downloads the OCR engine and that model. Later runs can reuse the cached files.
It adds searchable text, which helps. It does not add the tags, reading order or alt text that screen readers need. Check the result with the PDF Accessibility Checker.
Last reviewed . See how we build and test our tools, or report an error.
