OCR PDF
Read text from scanned PDFs and photographed pages with Tesseract, in your browser (on iPhone and iPad, on our own OCR service)
Click or drop a PDF file hereChoose a PDF file
Type an English name, a native name, or a code to search; choose up to 3 languages.
About OCR PDF
OCR PDF reads the text in pictures: each page of your PDF is drawn on a canvas and recognized by Tesseract, so scanned documents and photographed pages inside a PDF become text. Choose the language among 102, or up to three for a mixed document; your browser's language is picked first when it is in the list. You get the recognized text to copy and a searchable copy of your PDF: the original pages with an invisible text layer added. Pages that already had their own text do not get a second layer. The input must be a PDF; turn photos into one with Image to PDF first.
How to OCR a scanned PDF
- 1Choose the PDF to read.
- 2Type in "Language of the document" to search by English name, native name, or code, and choose one to three languages.
- 3Click "Run OCR". On a computer or Android device, the first run downloads the engine and the language data; on iPhone and iPad, the PDF goes to our OCR service. The pages are read one by one.
- 4Copy the result with "Copy text", or click "Download" for the -searchable.pdf file.
Formats and limits
- Input
- PDF only: scans or photos placed in a PDF
- Languages
- 102, up to three at once
- Output
- Recognized text, and a searchable PDF when text was found
- On iPhone and iPad
- Every page is recognized by our OCR service, for PDFs up to 44 MB
- Usage limits
- Our OCR service (iPhone and iPad only): 300 pages per hour and 1,000 pages per day per network
Where your file is processed
On a computer or an Android device, recognition runs in your browser with Tesseract.js and your PDF is not uploaded; the engine and language data are downloaded from cdn.jsdelivr.net the first time. On an iPhone or iPad, every page is recognized by our own OCR service (Tesseract on our pdf-tools server), because the recognition did not finish on a real iPhone in our tests: the PDF is sent there, then deleted, and the page tells you before and after.
Frequently asked questions
How accurate is the recognition?
0.8% of characters were wrong in our one measured test (September 29, 2026, text printed at 7 pt). Skewed, blurred, or low-contrast pages give more errors, and a wrong character can look plausible, such as an 8 read as a 3, so check names and figures.
Can I OCR a JPG or PNG photo directly?
No. The tool opens PDF files only. Put the photos into a PDF with Image to PDF or JPG to PDF first, then run OCR on that PDF; each photo becomes one page that is recognized like a scan.
Will the searchable PDF look different?
No. The pages are not redrawn or recompressed: the recognized text is added as an invisible layer placed on each page as it is displayed, crop and rotation included. A page that already had at least half as much selectable text is left without a second layer.
Is the first run slower?
On a computer or an Android device, yes: the OCR engine and the data of each chosen language are downloaded the first time, with their own progress bar. Your browser keeps them, so later runs only download a language you have not used before. On iPhone and iPad the OCR engine is not downloaded, since our OCR service reads the pages.
Is there a limit on iPhone or iPad?
Yes. Our OCR service, which reads the pages there, accepts PDFs up to 44 MB and recognizes at most 300 pages per hour and 1,000 pages per day per network. On a computer there is no fixed limit, but a step stuck for 90 seconds stops with a message.
Tips and tricks
- ✓If letters come out wrong, check the language first: a page written in two languages reads better with both selected.