OCR PDF: Make Any Scanned PDF Searchable
A scanned PDF is really just a photograph of a page: you can read it, but Ctrl+F cannot, and neither can a screen reader, a search engine or copy paste. This tool fixes that with OCR (optical character recognition). It reads every page, works out where each word sits, and builds a new PDF that looks identical to your scan but carries an invisible text layer underneath, so you can search it, highlight it and copy from it. Everything runs in your browser on your own device, so your document is never uploaded. An honest note before you start: OCR is slow and clear scans recognise best, so a long document takes minutes, not seconds.
Last updated: October 22, 2026
The formula
Render: each PDF page is drawn as an image at about 300 DPI; an uploaded scan image is used directly, and small images are upscaled first.Recognise: Tesseract OCR reads the rendered page in the language you picked and returns every recognised word with its exact position on the page.Text layer: a new PDF is built with the page image on top and every word drawn again underneath it as invisible text, at the same position and size.Result: the standard Searchable Image format, the page you see is your scan, and the words you search, highlight and copy live in the hidden layer.
A worked example
Worked example, from this tool: a one page scanned style invoice image (1700 × 2200 px) headed "INVOICE 1042", with "Total due: $1,250.00" and "Payment due on receipt". The OCR recognised all 30 words on the page, and the finished PDF's hidden text layer contains the full invoice text, searching the PDF for "1250" or "receipt" finds and highlights the words on the page, while the page itself still looks exactly like the original scan.
More about the OCR PDF
Why your scanned PDF is not searchable (and what OCR does about it)
When a scanner or a phone camera makes a PDF, the page is stored as one big picture. The words you see are shapes in that picture, there is no text in the file at all. That is why Ctrl+F finds nothing, why you cannot select a sentence, and why a screen reader reads the page as blank. A born digital PDF (exported from Word, a browser or a design tool) is the opposite: real text wearing a page layout, searchable from birth.
OCR, optical character recognition, bridges the gap. The engine looks at the page image, identifies the shapes as letters and words, and records what each word says and where it sits. This tool then writes those words back into the PDF as an invisible layer, positioned exactly over the scanned words. The file that comes out is the format archivists call a Searchable Image PDF: picture on the surface, real text underneath.
Getting the best recognition out of your scan
Clear scans recognise best, that one sentence is most of OCR. Straight pages, dark text on a plain light background, and a sharp image at a decent size give the engine easy work; this tool renders PDF pages at about 300 DPI and upscales small images for the same reason. Blurry, skewed, shadowed or very faint pages are where mistakes creep in: a 0 read as an O, an rn read as an m. If a page matters, rescan it straight and well lit rather than hoping software will rescue it.
Second, pick the right language. The language setting tells the engine which alphabet and dictionary to expect, and it is the single biggest accuracy lever after scan quality, running a Spanish document through the English model garbles every accented word. Third, proofread anything you will rely on. OCR output is a reading of your document, not the document itself: check totals, dates, names and reference numbers in the result before you file or send it.
What happens to your file, and the honest limits
Your file never leaves your device. The OCR engine and its language data download to your browser on first use (which is why the first run takes a little longer), and from then on the reading, the rebuilding and the download all happen locally. There is no server copy, no account and nothing to delete later, for confidential scans, that is the whole point.
The limits, without varnish. The output stores each page as a high quality image of the original page, so a born digital PDF with crisp vector text becomes an image page with a text layer, searchable again, but no longer vector. Handwriting is not read reliably; this is a printed text tool. Very long documents are slow and can strain a phone's memory, so split a book into chapters first. And for non Latin scripts such as Arabic and Urdu, recognition runs, but the text layer is written with the built in Latin font, so some recognised words cannot be stored in it. Latin script languages get the complete layer.
How to use the OCR PDF
- Click Select PDF or image file and choose a scanned PDF, or a PNG, JPG or WebP scan.
- Choose the language the document is written in, English is the default.
- Press Make Searchable PDF and wait: the first run downloads the OCR engine and language data, then every page is read in turn.
- Watch the progress line, it names the page being read and the overall percentage as it goes.
- When it finishes, check the summary of pages and words recognised, then click Download Searchable PDF to save it.
Frequently asked questions
What does "searchable PDF" actually mean?
It means the PDF contains a real text layer as well as the page picture. Viewers can find words with Ctrl+F, highlight matches, and let you select and copy text, even though what you see on screen is still the scanned page. A plain scanned PDF has only the picture, so none of that works until OCR adds the text layer.
Will making my PDF searchable change how it looks?
No. The visible page is your original scan, stored as a high quality image at the same page size. The recognised words are added as invisible text underneath it. You cannot see them, but search, highlighting, selection and copy paste all use them. If you zoom right in, the page is an image of the scan rather than crisp vector lines, exactly as the scan itself was.
How accurate is the OCR?
It depends almost entirely on your scan, so beware any tool that quotes you a percentage. Clean, straight, high contrast printed pages recognise very well, in our own build test, a crisp one page invoice was recognised word for word, all 30 words. Blurry, skewed, faint or crumpled pages produce mistakes, especially in numbers and names. Always proofread figures you plan to rely on.
Can it read handwriting?
Honestly, no, not reliably. This tool uses Tesseract, which is built for printed text: typed, typeset or clearly print like lettering. Handwriting usually comes back garbled or empty, and the tool will tell you when no text was recognised rather than inventing some. For handwritten notes, typing them out is still the dependable route.
Which languages can it recognise?
English, Spanish, French, German, Arabic and Urdu, pick the document's language in the dropdown before you convert, because the engine uses that language's alphabet and dictionary. One honest caveat: the invisible text layer is written with the built in Latin font, so for Arabic and Urdu some recognised words cannot be stored in the layer. Latin script languages get the complete text layer.
Is my document uploaded to a server?
No. The OCR engine and language data are downloaded to your browser once, and then your document is read and rebuilt entirely on your own device. Nothing is uploaded, stored or kept, close the tab and there is no copy anywhere. That makes this a safer choice for confidential scans than services that process files on their servers.
What is the difference between a scanned PDF and a born digital PDF?
A born digital PDF was exported from an app. Word, a browser, a design tool, and already contains real, searchable text. A scanned PDF is a photograph of paper: the words are part of the picture and there is no text to search. If your PDF already lets you select text with the mouse, you do not need OCR at all. If selection grabs the whole page as one image, OCR is the fix.
Can I copy and paste text from the finished PDF?
Yes, that is the point of the text layer. Open the downloaded PDF in any viewer, drag across the words, and copy: you get the recognised text. The pasted text follows the reading order of the words on the page and carries no formatting, so multi column pages can paste in a jumbled order. For a fully editable document, run the result through our PDF to Word tool afterwards.
Are there page or file size limits?
There is no artificial page cap, quota or daily limit. The real limit is your device: every page is rendered and read in your browser's memory, so a very long or very large scan can slow down or run out of memory on a phone or an old laptop. For a book length document, split it into chapters with our Split PDF tool, OCR each part, then merge them back together.
How can I tell if my PDF is already searchable?
Two quick tests. First, try to select a single word with your mouse or finger: if the whole page highlights as one block, the page is an image with no text layer. Second, press Ctrl+F (Cmd+F on a Mac) and search for a word you can clearly see on the page, if the viewer finds nothing, the PDF is not searchable yet and OCR is the fix. If both tests pass, your PDF already has text and does not need this tool.