How to Fix a PDF Where Text Is Not Selectable (OCR Fix)
You opened a PDF and tried to copy a paragraph — but instead of text, your cursor highlights nothing. The PDF is an image-only scan, not real text. OCR fixes this in one step.
The problem
A scanned contract was sent to you as a PDF. You need to copy a clause for a reply email. Selecting text doesn't work because the "text" is just an image of text. Without OCR, you'd retype it character by character.
Use the tool now
Open the ocr pdf tool and follow the steps below.
Step-by-step
- 1
Confirm it's an image-only PDF
Try selecting text. If your cursor doesn't highlight letters, it's a scan. Look at zoom level — image-only PDFs become pixelated when zoomed; text PDFs stay sharp.
- 2
Open the OCR PDF tool
Drop the PDF in. PDFShed runs Tesseract OCR locally in your browser.
- 3
Pick the document language
Tesseract supports 100+ languages. Pick the right one — English OCR on a French scan returns garbage.
- 4
Wait for OCR processing
A 20-page scan takes 30-60 seconds. Each page is analyzed for text and a hidden text layer is added underneath the image.
- 5
Download and verify
Open the new PDF and try selecting text. Now it works. The visual scan is preserved; selectable text is overlaid behind it.
Pro tips
- •OCR accuracy depends on scan quality. Crisp 300 DPI scans get 99% accuracy. Faxed or low-quality scans drop to 80-90%.
- •After OCR, the file size grows ~10-20% from the added text layer. Run [Compress PDF](/en/tools/compress-pdf) if size matters.
- •For multi-language documents (e.g., English + Chinese), pick the dominant language. Mixed-language OCR is roadmapped.
- •OCR text is searchable in Acrobat, Preview, and any modern reader after this conversion.
Frequently asked questions
Why was my PDF not selectable in the first place?
It was a scanned image — created by a scanner or "Print to PDF" of an image-based source. The "text" is part of the image, not real text.
Will the visual layout change after OCR?
No — the original scan stays exactly as-is. OCR adds an invisible text layer behind it for selection/search. Visual fidelity 100% preserved.
How accurate is OCR for handwriting?
Tesseract is poor at handwriting (~50-70% accuracy). For handwritten content, manual transcription is faster than cleanup.
Can I OCR a 500-page document?
Yes, but it takes a few minutes. Browser RAM is the limit — close other tabs first.