Document language
Choose the language matching the scan.
A mismatched model reduces recognition accuracy.OCR PDF
Turn scanned pages into text you can search, select, and copy.
Drop your PDF file here
Choose a file to get started.
Files stay on your device No sign-up Free to use
How it works
Open the scan to recognize
Select your scanned PDF.
Match the language and text output
Choose its language and whether to include a text file.
Proofread recognized content
Run OCR, wait for each page to finish and review the result.
SCANNED PAGEMeeting notes
Ideas for the next project.
Pick up where we left off.
RECOGNIZED TEXTMeeting notes
Ideas for the next project.
Search it. Select it. Copy it.Optimize
Your scans, now searchable.
Searchable archive
An English scan contains invoice references needed for search. Select English and request the text file for checking.
✓ Search for a known reference, copy it and compare every digit with the scan.Make it work for your document
Options after selecting a fileText file
Adds a plain-text result alongside the OCR PDF.
Text loses page layout and requires proofreading.Frequently asked questions
The details, when you need them.
Does OCR work without an upload?
Recognition runs in your browser. Software and language models are downloaded, but this tool does not send your PDF to a processing server.
Can OCR guarantee accurate text?
No. Always review the recognized text, especially numbers and low-quality scans.
Improve recognition before you start
Straight pages with sharp text and even contrast give recognition software a better starting point. Blurred photographs, handwriting, complex columns and small type can produce mistakes. Select the language that matches the document rather than leaving the default when the text is in another language. Proofread names, numbers and other important details.
Allow time for large scans
The first run downloads OCR software and a language model. Recognition then uses your own device, so a long scan can take substantial time or memory. Keep the tab open and avoid running several large jobs at once. Searchable output is useful for finding passages, but it does not guarantee a perfect reconstruction of the original reading order.
Recognition requires proofreading
Recognition can misread names, digits, punctuation and reading order.
Text is incorrect despite readable pages
Low resolution, skew or a mismatched language can confuse recognition. Use a clearer scan and the correct language; proofread the result.