PDFApproximateRuns in your browser

OCR PDF

Read text out of scanned documents and photos with on-device OCR. The result is a best-effort reconstruction; check it before relying on it.

Not uploaded to any serverConverted on this deviceGone when you close the tabHow this works

Drop your files here

PDF, JPG, PNG, WebP and 1 more formats · up to 100 MB each

or paste files from your clipboard

Ready when you are

Read text out of scanned documents and photos with on-device OCR. Everything happens in this tab; nothing is sent anywhere.

⌘K to search converters

How to use it

1
Add your files
OCR PDF works with PDF, JPG, PNG, WebP and TIFF. Drop them in, browse for them, or paste from your clipboard.
2
Choose your options
Tune the output settings — or leave the defaults, which are sensible.
3
Run it and download
Press the button and save the result. Everything happens on your device, and nothing is uploaded.

Privacy, in plain words

Is it safe? Your PDF is not uploaded anywhere

What “uploaded” normally means. On most online converters, pressing the button copies your file across the internet onto a computer owned by that website. That copy is what the word “upload” means, and it is the part nobody explains.

What happens instead. This site does not do it. There is no server here that receives files, so your PDF is read, converted and handed back by the browser you are using right now. It never travels — so there is no copy to store, read, leak or hand over.

Check it yourself in ten seconds

Turn off your Wi-Fi after this page has loaded, then convert a file anyway. It still works, because everything needed is already on your device.

The longer explanation of how this works, and the full detail of what is and is not stored on the privacy page.

Why this matters more than it sounds

PDF is the format people use for exactly the documents they would not hand to a stranger. It matters more than it sounds, because of what PDFs usually are:

Medical letters and results

a diagnosis is not something to leave on a server

Payslips

salary, employer, and usually your home address

Bank statements

account numbers, balances, every payee you have

Signed contracts

your signature, and terms you may be under NDA about

A file that is never sent cannot be kept, leaked, subpoenaed or used to train anything. That is a property of how this works, not a promise about how carefully it is run.

What recognition can and cannot do

A scanned PDF contains pictures of words, not words. Nothing can select them, search them or copy them, because as far as the file is concerned each page is a photograph. Recognition looks at those pictures and works out which letters they show, then writes that text back into the document underneath the image.

It is a reading of your page rather than a copy of it. Clean printed text on a straight scan is recognised very accurately; handwriting, faint photocopies, tight columns and unusual typefaces are where mistakes appear. Always read the output before relying on it.

The page still looks the same: The original image stays on top. The recognised text sits behind it, so the document reads exactly as before but is now searchable.
Language matters: Recognition uses a model per language. Choosing the wrong one produces confident nonsense rather than an error.
Quality in, quality out: A 300 DPI scan recognises far better than a photograph of a page taken at an angle.

Check whether you need this at all

Most PDFs are not scans. If yours was produced by Word, a browser or any ordinary program, the text is already inside it and extracting it is instant and exact — no recognition, no mistakes, no waiting.

The quickest test is to open the file and try to select a line of text with your cursor. If it highlights, you do not need recognition and should use extraction instead. If the whole page highlights as one block, it is a scan and this is the right tool.

About these formats

Accepts

Portable Document Format

Fixed-layout documents that look identical everywhere.

Extensions: .pdf
MIME type: application/pdf
Best for: Sharing documents · Printing · Contracts and forms
Compatibility: Universal.

Accepts

JPEG Image

The universal lossy photo format.

Extensions: .jpg, .jpeg, .jpe, .jfif
MIME type: image/jpeg
Best for: Photographs · Email and messaging · Maximum compatibility
Compatibility: Opens everywhere — every browser, OS and device.

The same tool as Extract Text from PDF: a page that already contains text is read exactly rather than recognised, and only images go through OCR. Runs entirely in your browser in any of 66 languages. Handwriting and skewed photos are not reliably recognisable.

Good to know

OCR PDF — questions

How does OCR PDF work?

Read text out of scanned documents and photos with on-device OCR. Add your files above, adjust anything you want to change, and press the button. The result downloads straight to your device.

Is there anything I should watch out for?

Is OCR PDF safe — are my files uploaded to a server?

No, nothing is uploaded. On most online tools, pressing the button copies your file across the internet onto a computer owned by that website, where it is processed and then sent back. That copy is what "upload" means. This tool does not do it: your file is read and processed by the browser you are using right now, stays on this device the whole time, and is discarded when you close the tab. Turn off your Wi-Fi after the page loads and it still works, which is the quickest way to confirm it.

Can I use this on confidential or personal documents?

Yes — that is the case it was built for. Because the file never leaves your device, a confidential document is no more exposed than it is sitting in your own folder: there is no transfer, no copy on anybody else's hardware, no retention period and no staff with access. Nothing about the file is logged either — not its name, not its size, not its contents.