PDF OCR
Read the words on scanned pages, compare them with the page image and fix what was misread. The result can be a searchable PDF or a plain text file.
Add one PDF, scanned or not, to read its text.
or drop a file here
- Pages ready
- 0
- Words
- 0
- Characters
- 0
Save the result
The searchable PDF keeps every page exactly as it is and adds the recognized words as invisible text, so you can search, select and copy them. Your corrections are included.
Your PDF is handled inside this browser tab and is never uploaded.
How to use it
- Press ‘Choose a file’ or drop one PDF on the tool. Its pages appear as a strip of thumbnails, each marked ‘Scanned’ or ‘Text layer’.
- Under ‘Text recognition languages’, select every language the document uses. English is selected to begin with, and your choice is remembered.
- For a long document, choose ‘Page range’ under ‘Pages’, type something like 1-3, 5 and press ‘Apply’. This option appears only when the file has more than one page.
- Press ‘Recognize all pages’, or open one page and press ‘Recognize this page’. Pages that already hold real text are read directly.
- Compare the page with its text and correct anything misread. On a wide screen the page is on the left and the text on the right. ‘Undo my edits’ brings back the recognized text of that page.
- Under ‘Save the result’, press ‘Searchable PDF’ for a copy with an invisible text layer, ‘Text file (.txt)’ for plain text, or ‘Copy all text’ to paste it elsewhere.
What it does
Every page next to its text
The page image and its text are shown together, and the text sits in a box you can edit. ‘Show word boxes’ tints each recognized word on the page, so you can see what was read where.
A searchable copy that looks unchanged
‘Searchable PDF’ keeps every page exactly as it is and lays the recognized words over it as invisible text, so a search finds them and you can select or copy them. Your corrections go in too, and the file is named like the original with -ocr added.
Pages that already have text
A page that already holds real text is read directly and labeled ‘Text layer’, with no recognition needed. If that text looks wrong, ‘Recognize anyway’ reads the picture of the page instead.
Seven recognition languages
Pick any of English, French, German, Japanese, Korean, Portuguese and Spanish, alone or mixed on the same page. Each is labeled with its own name, such as Deutsch or 한국어. Fewer languages recognize faster, so select only the ones in your document.
Search across every page
‘Search all pages’ counts the matches in the whole text, tints them on the page and lists the pages they are on, such as p. 3 (2). The counters above show pages ready, words and characters.
Recognized on your device
The text is read by Tesseract OCR running in your browser, so the PDF is not uploaded. The first run downloads the language data from this site, and your browser keeps it for next time.
Good to know
A scanned lease, a receipt photographed into a PDF or an old report is just a picture on each page, so it cannot be searched or copied. Once it is recognized you can find a clause by its words, paste a paragraph into an email, or file the document so that a search turns it up. Use ‘Text file (.txt)’ when you only need the words, and ‘Searchable PDF’ when the pages must stay as they are.
Recognition works best on straight, sharp scans of dark print on light paper. Small type, stamps, handwriting, folds and faint copies are misread more often, so check names, dates and amounts against the page before you rely on them. The text file marks each page of a longer document with a line like — Page 2 —, and the corrections you type go into both files.
Scanned pages are read one at a time, so a long scan takes a while, more so on a phone. ‘Stop’ ends the run and keeps the pages already done. A PDF that needs a password to open cannot be read, so remove the password with Unlock PDF first. For a quick text file without checking page by page, PDF to Text also reads scanned pages.
Questions and answers
How can I search the text in a scanned PDF?
Add the PDF and press ‘Recognize all pages’. Look the text over, then press ‘Searchable PDF’. The file you get looks like the scan, but its words can be searched and copied.
Will my pages look different afterwards?
No. Each original page stays as it was. The recognized words are laid over it as invisible text, each one where it appears on the page.
Can it read Korean or Japanese?
Yes. Korean (한국어) and Japanese (日本語) are among the seven languages. Select them under ‘Text recognition languages’ before you start, together with any other language in the document.
Why are some words wrong?
Faint print, small type, stamps and handwriting trip up recognition. Fix the word in the text box beside the page, and both saved files will use your fix.
Does it work on a PDF that already has text?
Yes. A page with its own text has that text taken straight from the file, and saving a searchable PDF leaves the page as it is. The message under the buttons says how many pages received a new text layer.
Is my PDF uploaded?
No. Recognition runs in your browser, so the PDF stays on your device. The only download is the language data, which comes from this site the first time you run recognition.