OCR and text tools for scanned PDFs
A scanned PDF looks like a document but is really a stack of pictures, so you cannot search it, select words or convert it to Word. OCR reads those pictures and produces real text. Here is how to tell whether you need it, and which related tools to use before and after.
1 tools
Scanned PDF or text PDF: how to tell
Open the file and try to select a sentence with your mouse, or search for a word you can see on the page. If words highlight and the search finds them, the PDF already has a text layer and does not need OCR. If you can only select the whole page as one block, or nothing is found, it is a scan made of images.
Documents from scanners, phone camera photos and files saved as images then turned into PDF are all scans. Exports from Word or web pages are usually text PDFs. Some files mix both, with a few scanned pages inside a text document, which is why OCR PDF has an option to skip pages that already contain text.
What OCR PDF produces and how to get good results
OCR PDF renders each page to an image, recognizes the characters with Tesseract.js running in your browser, and outputs a searchable PDF (the page image with an invisible text layer) and, if you like, a plain text file. It reads English, French, Spanish, Arabic, German, Italian, Portuguese, Dutch, Turkish and Russian, and you can choose several at once for mixed documents. It also accepts JPG and PNG images directly.
Accuracy depends mostly on the scan. Aim for 300 DPI, straight pages, good contrast and clean lighting; a crooked, shadowed phone photo will give more errors. Handwriting is not reliably recognized. For Arabic and Russian, the tool provides the recognized text as a .txt download because the searchable-PDF layer is not available for those scripts. Expect seconds per page on a laptop and longer on a phone.
- Choose every language that appears in the document, not just the main one.
- Use 300 DPI for small print and 200 DPI for quick, clean pages.
- Rescan a poor page rather than expecting OCR to fix it.
- Proofread names, numbers and dates, where a single wrong character matters.
Tools to use before and after OCR
Before OCR, Scan to PDF or Images to PDF can assemble photographs into a single PDF, and Rotate PDF fixes pages that were scanned sideways, because OCR reads upright text best. After OCR, PDF to Text extracts the text layer to a .txt file, PDF to Word rebuilds an editable document and PDF to Markdown gives structured plain text. Compare PDF finds differences between two versions, but like the others it works on real text, so scanned files need OCR first.
OCR PDF runs in your browser and your documents are never uploaded. The OCR engine and language data, a few MB, are downloaded once from a public CDN (jsDelivr) the first time you use it; the CDN sees your IP address but not your documents.
Frequently asked questions
How do I know if my PDF needs OCR?
Try selecting text or searching for a word you can see. If nothing highlights or the search finds nothing, the PDF is a scan and needs OCR. If text selects normally, it already has a text layer.
Are my documents uploaded when I run OCR?
No. Recognition runs in your browser. The first time you use it, the OCR engine and language files are downloaded from a public CDN (jsDelivr), which sees your IP address but never your documents.
Does OCR work for Arabic and handwriting?
Arabic is recognized, but the output is provided as a .txt download because the searchable-PDF text layer is not available for Arabic or Russian. Handwriting is not reliably recognized; OCR works best on clean printed text.
Why is OCR slow on my file?
Each page is rendered as an image and analyzed on your own device, which takes seconds per page on a laptop and longer on a phone. For a large document, use a computer, choose 200 DPI if the print is clear and keep the tab open.
What should I do after OCR to edit the text?
Use PDF to Word for an editable document, or PDF to Text for plain text. Proofread the result first, because OCR can misread characters, especially in poor scans.