Convert Photo to Text
Extract editable text from a photo — pick a language, run OCR in your browser, then copy or download the result.
Drop an image to read
Drag & drop, paste, or pick a file
PNG · JPG · WebP — text is read on your device
Turn a photo into editable text
Snapped a photo of a page, a whiteboard, a menu, or a business card? This page turns that photo into text you can edit. Add the image, pick the language, and the tool reads it right on your device — no typing, no re-keying.
Photos are the trickiest input for OCR because of angle, lighting, and focus. Two quick habits fix most problems: shoot the page flat and straight-on rather than at an angle, and make sure it's evenly lit with no glare or deep shadow. A sharp, upright photo reads almost as well as a scan.
Best for
- Photos of book or document pages
- Whiteboards and lecture notes
- Business cards and menus
- Product labels and packaging
How it works
- Add your image — drag & drop, paste, or choose a file.
- Pick the language of the text for the best accuracy.
- Press Extract text and watch the progress bar.
- Edit the result, then copy it or download a .txt or searchable PDF.
What this does & how to get the cleanest text
- What it does
- Turns a photo of a page, whiteboard, menu, label, or card into editable text, recognised on your own device with Tesseract.js.
- What you get
- An editable text box (copy it, download .txt, or export a searchable PDF) — handy for re-keying a page you only have on camera.
- Runs in
- Your browser, on your device (Tesseract.js) — the engine and your chosen language download once, then it works offline.
Tips for accurate photo recognition
- A photo is the hardest input for OCR because angle, lighting, and focus all vary at once — shooting the page flat and straight-on, filling the frame with it, fixes most accuracy problems on its own.
- Tap to focus before the shot: a slightly blurry photo is the single most common reason recognition comes back garbled, and it's the easiest to avoid.
- Avoid glare and deep shadow; diffuse, even light across the whole page reads far better than a bright spot with a dark corner.
- Crop tight to just the text area if there's a lot of background — fewer non-text pixels means fewer stray characters in the result.
- Printed signage and typeset pages read well; handwritten notes on a whiteboard are recognised only roughly, so expect to do more correcting there.
Languages & scripts it can read
Pick the language that matches your image before running OCR — it loads the right model and improves accuracy, and it's required for non-Latin scripts. The curated selector covers these scripts:
| Script | Example languages |
|---|---|
| Latin | English, Spanish, French, German, Vietnamese |
| Cyrillic | Russian, Ukrainian |
| Greek | Greek |
| Arabic | Arabic, Persian, Urdu |
| Hebrew | Hebrew |
| Devanagari | Hindi |
| Other Indic | Bengali, Tamil, Telugu |
| Thai | Thai |
| CJK | Chinese (Simplified/Traditional), Japanese, Korean |
Arabic, Persian, Urdu, and Hebrew are read right-to-left and the result box flips to match. Printed and typed text is recognised most reliably; handwriting and very small or decorative type are less accurate, so review the result.
Frequently asked questions
Is it free?
Yes — completely free, with no watermark and no sign-up.
Where does the text recognition happen?
It happens in your browser, on your own device. The first run downloads the OCR engine and your chosen language; after that it works offline. How any data associated with the tool is handled is described in our privacy policy.
Which languages can it read?
Over 100, across Latin, Cyrillic, Arabic, Hebrew, Devanagari and CJK scripts. Choose the one that matches your image before running OCR.
Can I get a searchable PDF?
Yes — after extracting, choose Download PDF for the original image with a selectable text layer on top.