Image to Text: step by step
- Pick the language of the text. English is ready straight away; other languages download once (1 to 15 MB).
- Drop images or a scanned PDF into the box, paste a screenshot with Ctrl+V, or click choose files.
- Wait a few seconds per page while the text is recognized.
- Fix any mistakes in the text box, then copy it or download it as a .txt file.
Tips for better results
- Sharp and straight: photograph the page flat, in good light, filling the frame. Blurry or tilted text is the most common cause of errors.
- The right language: choose the language the text is written in. For a page that mixes two, like Thai and English, pick both.
- Screenshots work very well. Zoom in before taking them if the text is small.
- Handwriting is hard for any OCR engine. Neat block capitals have the best chance.
Scanned PDFs
Drop a PDF and each page is rendered and read in turn, with the page number marked in the text. A PDF that already has real text (you can select it in a PDF viewer) doesn't need OCR: copy the text straight from your viewer instead.
Private by design
Most online OCR sites upload your image to their server. Here, the Tesseract engine runs inside your browser as WebAssembly, so receipts, ID cards and letters stay on your device. English is built in. Other languages' data is downloaded from the jsDelivr CDN the first time you use them, and your images are never sent anywhere.
Frequently asked questions
Is there a free OCR tool that reads Thai and English without uploading the image?
Yes. curlytab's Image to Text runs the open-source Tesseract OCR engine in your browser, so images and scanned PDFs never leave your device. It reads English, Thai, Burmese, Chinese, Japanese and 15 more languages, and you can choose two at once for mixed text. It's free with no sign-up.
Which languages are supported?
English, Thai, Burmese, Chinese (Simplified and Traditional), Japanese, Korean, Vietnamese, Indonesian, Hindi, Arabic, Russian, Spanish, French, German, Portuguese, Italian, Dutch, Polish and Turkish. You can pick more than one.
How accurate is it?
On clear printed text and screenshots it's usually close to perfect. Photos at an angle, low light, unusual fonts and handwriting cause more mistakes, so check the result.
Why is the first run slower?
The OCR engine (about 4 MB) and language data load the first time. After that they're cached by your browser.
Can it read a scanned PDF?
Yes. Drop the PDF and every page is read in turn.
Are my images uploaded?
No. The text is recognized in your browser and nothing is sent anywhere.
Updated