No text found. Try cropping, rotating or choosing another language.
OCR can misread characters. Check the text before you rely on it.
Convert photos and scans of Russian text into editable Cyrillic, with English words recognized in the same pass. Russian + English is selected from the start; keep the text at least 20 px tall and straight for the cleanest result, and check words marked in yellow.
How to use
- Drop images or a PDF, paste a screenshot with Ctrl+V, or take a photo with Camera. Several files go into a queue.
- Pick the language of the text. Combine languages (for example English + Japanese) for mixed text.
- Text appears as soon as each image is read. Edit it in place, then copy it or save it as a .txt file.
- If the result looks off, drag on the image to read only one area, rotate it, or toggle Grayscale, Contrast or Invert. The image is read again automatically.
- Open Check to see uncertain words highlighted, and the boxes on the image show where they are.
- Recognition runs on your device with Tesseract. Images are never uploaded. The language data (1–3 MB each) is downloaded once and cached.
- Best results come from sharp, straight text at least 20 pixels tall. Small screenshots are enlarged automatically, and dark-mode screenshots are inverted.
- Handwriting, decorative fonts and text over busy photos are hard for OCR. Use Check to fix the words it was unsure about.
- Confidence is Tesseract’s own estimate per word: green is reliable, yellow worth a look, red likely wrong.
How the OCR works
Recognition uses Tesseract, an open-source OCR engine, compiled to run in the browser. The language data for each language (about 1 to 3 MB) is downloaded once and then cached, so later images start right away. Combine languages, such as English + Japanese, for mixed text. Each word gets a confidence score: green is reliable, yellow is worth a look and red is probably wrong.
Getting accurate results
OCR works best on sharp, straight, evenly lit text at least 20 pixels tall. Small screenshots are enlarged automatically and dark-mode screenshots are inverted. If the result looks off, drag on the image to read only one area, rotate a sideways photo, or try Grayscale, Contrast or Invert; the image is read again automatically.
Handwriting, decorative fonts, very low resolution and text over busy photos are hard for any OCR engine. Check the highlighted words before you rely on the text.
Privacy and PDFs
Images and PDF pages never leave your device, which makes the tool suitable for receipts, contracts and personal documents. Multi-page PDFs are rendered page by page and each page joins the queue. For PDFs that already contain selectable text, copying the text directly from a PDF reader is faster and exact.
Frequently asked questions
How do I copy text from an image?
Drop the image or paste a screenshot with Ctrl+V. The text appears as soon as it is recognized; edit it if needed and tap copy or save it as a .txt file.
Which languages are supported?
English, Spanish, Portuguese, French, German, Japanese, Russian and Korean. You can combine several, for example Korean + English for mixed text.
Can it read handwriting?
Only neat, printed-style handwriting with limited accuracy. Tesseract is trained on printed text, so cursive and messy notes usually come out wrong.
Can I extract text from a PDF?
Yes. Drop a PDF and each page is rendered as an image and recognized. If the PDF already has selectable text, copying it from a PDF reader is more accurate.
Why is the result inaccurate?
Common causes are small or blurry text, a tilted photo, low contrast or the wrong language. Crop to the text, rotate, enable Contrast or Grayscale, and check the language setting.
Are my images uploaded?
No. Recognition runs in your browser; only the language data is downloaded once, and your images stay on your device.