Extract text from an image (OCR)
A photo of a letter, a screenshot, a poster: get the text it contains, ready to paste into an e-mail, a Word document or a form.
Only with the 24-hour pass and Premium
or drag them here
Processed on your device, nothing is uploaded
Your file
The first time, the reading engine (about 5 MB) is downloaded and then kept on your device. Your pages, however, are never uploaded.
How do I extract text from an image (OCR)?
- Choose one or more images: JPG, PNG or photos from your phone.
- Choose the language of the text, then “Plain text (.txt)” as the output.
- Press “Recognize text”, then open the .txt file and copy what you need.
In an image, text is just a drawing of letters: it can't be selected. Optical character recognition (OCR) picks out these letters and turns them into real text. TwoDocument uses Tesseract, a leading open-source engine, which runs entirely in your browser here.
The result depends mainly on the image: printed text that is sharp, straight, well lit and large enough is recognized well, accents included. A blurry or tilted photo, or one taken from a distance, produces errors, and handwriting is not recognized reliably. Always check the text you get, especially figures: amounts, numbers, dates.
With several images, their texts are placed one after the other, in the order of the list. Choose “PDF with text” instead to keep the image and be able to search it for a word. The first time, the reading engine (about 5 MB) is downloaded from TwoDocument, then kept on your device; your images, however, are never uploaded.
Frequently asked questions
Which languages are recognized?
French and English, separately or together: choose “French and English” for a document that mixes the two.
Is handwriting recognized?
Not reliably: the engine is designed for printed text.
How can I get a better result?
Photograph the text flat, straight on, in good light and without shadows, and frame it as tightly as possible. For a whole document, run the photo through the Scan tool first, then run OCR on the resulting PDF.
Are my images uploaded to a server?
No. The recognition is computed by your device: even a photo of an ID document or a letter doesn't leave your phone or computer.
Other tools
Reopen a file
These files are kept on this device for 7 days.
You have used your 3 free documents for today
Come back tomorrow, or carry on right now:
24-hour pass: €1.99, no subscription Unlimited Premium: €6.99 a monthOnly with the 24-hour pass and Premium
This feature is included in the 24-hour pass and in Premium.
24-hour pass: €1.99, no subscription Premium: €6.99 a month