OCR a PDF online: make a scanned PDF searchable

OCR a PDF online by dropping the scan here: press OCR on the toolbar, wait while the text is recognized in your browser, and press Download to get a searchable PDF. The page images stay as they are; an invisible text layer is added so Ctrl+F and copy-paste work. Nothing is uploaded.

Drop a scanned PDF to make it searchableor click to choose. It opens in the viewer on this site, nothing is uploaded.Free, no account, no watermark. The file stays in this browser tab.

How to make a scanned PDF searchable

Drop the scanned PDF here or click Open a PDF.
Click OCR on the toolbar. The first time, the recognition engine and its language data download from a CDN; the PDF itself does not go anywhere.
Wait. Recognition runs page by page in the browser tab and takes tens of seconds per page on a laptop, so a 40-page scan is a coffee break.
Test it: press Ctrl+F and search for a word you can see on the page. Select some text and copy it.
Press Download. The new PDF has the invisible text layer embedded and looks exactly like the scan.

How OCR in the browser works and why it is slow

A scanned PDF is a stack of pictures. OCR (optical character recognition) looks at each picture, finds the letter shapes and turns them into text. This viewer uses tesseract.js, which is the Tesseract OCR engine compiled to run inside the browser. That is why nothing is uploaded: the recognition happens on your own CPU. It is also why it is slow compared with a server farm; a laptop core does the work that a cloud OCR service spreads across many machines.

The result is written as an invisible text layer placed over the image, in the same positions as the printed words. Readers show the picture, search and copy use the hidden text. The what is OCR page goes deeper; what is a scanned PDF explains how to tell a scan from a text PDF.

Languages

The OCR language is picked from your browser's language setting, with English as the default. A browser set to Russian recognizes Cyrillic; a browser set to English recognizes Latin text. If a document is in a language other than the one your browser uses, the recognition will be poor, since the engine is looking for the wrong alphabet. Language data is downloaded once per language from the CDN and reused.

What the result looks like in Acrobat, Preview and Chrome

The downloaded PDF behaves like any other searchable PDF. Adobe Acrobat Reader, Preview on a Mac, Microsoft Edge, Chrome, Google Drive and phone readers show the original scan and let you search, select and copy the recognized text. The text layer is invisible, so the page looks identical to the scan; what changes is that Ctrl+F finds things and the file becomes indexable by desktop search and document systems.

The viewer embeds the layer with pdf-lib when you download. Annotations you add in the same session, such as highlights on the newly recognized text, are saved in the same file.

Limits and gotchas

Accuracy follows scan quality. Clean 300 dpi scans of printed text recognize well; faxes, skewed photos, tiny type and handwriting do not. OCR cannot invent what the scanner did not capture.
Time. Tens of seconds per page on a laptop, longer on a phone. Hundreds of pages can exhaust a phone's memory; use a computer for big scans.
The scan still looks like a scan. OCR adds searchable text; it does not straighten, clean or sharpen the image.
No export to Word or text. The viewer does not convert formats. After OCR you can copy text out, or hand the searchable PDF to a converter; see PDF to Word and PDF to text.
Password-protected scans that ask for a password to open cannot be opened here.

Alternatives for OCR

Adobe Acrobat Pro: Recognize Text is fast and accurate, with many languages; the free Reader does not OCR.
Preview on macOS: no OCR for PDFs, though Live Text lets you select text in images on recent versions.
Microsoft Edge and Chrome: no OCR in the built-in viewers.
Google Drive: open a scanned PDF with Google Docs and Drive runs OCR on the server, producing an editable document.
PDF-XChange Editor (Windows): built-in OCR with a searchable-PDF output option.
Xodo: OCR in some of its apps and web tools.
Smallpdf, iLovePDF: fast server-side OCR; the scan is uploaded to their servers, which is the trade-off against browser OCR here.

Frequently asked questions

How do I make a PDF searchable for free?

Open the scan here, press OCR, wait for recognition, then download. The new file has an invisible text layer that search and copy use.

Does OCR change how the PDF looks?

No. The page image is untouched; the recognized text is added as an invisible layer underneath, so the file looks exactly like the scan.

Why is OCR taking so long?

Recognition runs on your own device in the browser rather than on a server. Expect tens of seconds per page on a laptop; long scans take minutes.

Can I OCR a PDF on a Chromebook?

Yes. The viewer runs in Chrome on ChromeOS and does the OCR locally, so nothing leaves the Chromebook.

Can I copy text from a scanned PDF after OCR?

Yes. Select the text on the page and copy; the invisible layer supplies the characters. Accuracy depends on scan quality.

Is my scan uploaded anywhere?

No. The OCR engine and language data are downloaded from a CDN once, but the document stays in your browser tab and is never sent to a server.