At a glance
- Accepts
- PNG, JPG, WebP, GIF and other images your browser decodes
- Outputs
- Editable plain text you can copy or download as .txt, with Tesseract's confidence score
- Where it runs
- Entirely in your browser, on your device
- What leaves your device
- Nothing - the file never leaves your device
- Limits
- Engine (about 1.4 MB) and each language (1-3 MB) download from the jsDelivr CDN on first use; printed text works best, handwriting poorly; no layout or tables
- Account
- Not required - there is no sign-up
- Price
- Free, with no watermark and no usage cap
- Made by
- Visrut
- Last updated
- 12 October 2026
This pulls the text out of an image so you can copy, edit or search it: a screenshot of an error message, a photo of a page, a scanned letter, a slide. It runs Tesseract, the open-source OCR engine originally developed at HP and later by Google, compiled to WebAssembly and running in a worker inside your browser tab. The image stays on your device.
How to use it
- Pick the language the text is written in. Each language downloads once, the first time you use it.
- Paste a screenshot, drop an image, or click to choose one.
- The text appears on the right. Fix anything it misread, then copy it or download it as a .txt file.
Getting a clean result
- Screenshots beat photos. If you can screenshot the text instead of photographing a screen, do.
- For a photo of paper, shoot straight on in good light and crop to the text. Skew, shadows and blur cost the most accuracy.
- Tiny text is hard to read for software as it is for people - zoom in before taking the screenshot.
- Handwriting is mostly out of reach. Tesseract is built for printed type.
Frequently asked questions
Is my image uploaded to a server?
No. The text is recognised by Tesseract running in your browser tab. The only downloads are the OCR engine (about 1.4 MB) and the language you pick (1 to 3 MB), both fetched from the jsDelivr CDN the first time and cached afterwards. Your image is never sent anywhere.
What kinds of images work best?
Screenshots and clean scans of printed text work best: sharp, straight, dark text on a light background, at least 20 pixels tall per line. Phone photos of documents work if they are in focus and not too skewed. Handwriting, stylised fonts, and text over busy backgrounds come out poorly.
What does the confidence number mean?
It is Tesseract's own estimate of how sure it is, averaged over the words it found. It is not a measured accuracy, and a confident engine can still be wrong, but a low number reliably means you should proofread carefully.
Which languages are supported?
English, Spanish, French, German, Portuguese, Italian, Dutch, Hindi, Russian, Arabic, simplified Chinese, Japanese and Korean. Each language is a separate download of roughly 1 to 3 MB, fetched only when you pick it. Choose the language the text is written in - English letters read with the Hindi model, for example, come out as nonsense.
Does it keep the layout, tables or formatting?
No. You get plain text with the line breaks Tesseract found. Columns are read one after another, and tables come out as rows of words without the grid.
Can it read a PDF?
Not directly. Convert the page to an image first with the PDF to PNG tool, then drop the image here.