PDF Labs
OCR PDF: Make Scans Searchable
Make a scanned PDF searchable and copyable with OCR that runs in your browser. Adds an invisible text layer, or saves a text file. English only.
Drop the scanned PDF
The first run loads about 7 MB of recognition files from this site; after that it works offline.
How to use OCR PDF: Make Scans Searchable
-
Open the scan
Drop the scanned PDF.
-
Choose the result
Pick Searchable PDF or a text file, and a resolution.
-
Read and download
Press the button and download the result. The first run loads the recognition files.
About this tool
Turn a scanned PDF into one you can search and copy from. Each page is read with Tesseract, the open-source text recognition engine, running as WebAssembly in your browser. The recognised words are placed over the page as invisible text, so the PDF looks exactly as it did but Ctrl+F finds words and you can select and copy them. You can also save just the text.
Nothing is uploaded: the recognition engine and its English data are served from this site, loaded the first time you use the tool (about 7 MB) and kept for offline use afterwards. Pages that already have text are skipped, so a mixed document is only read where it needs to be.
This is a Labs tool. Recognition is English only for now, and it is not perfect: poor scans, handwriting, unusual fonts and small print produce mistakes, so check important numbers and names against the page. Higher resolution settings help with small type, at the cost of speed. Scans that are crooked or low-contrast read worse.
How do I make a scanned PDF searchable?
Open it here and choose “Make searchable”. Each page is read, and an invisible text layer is added so you can search and copy from the PDF, while the pages look unchanged.
Is my document uploaded for OCR?
No. Recognition runs in your browser using files served from this site. The first use loads the engine and English data; after that it works offline.
Why are some words wrong?
OCR makes mistakes on poor scans, small print, handwriting and decorative fonts. Try 300 dpi for small type and check anything important against the page.
Which languages are supported?
English only at the moment. Other languages need their own data files, which may be added later.