Turn an image into editable text

Extract printed English from a screenshot or scan, check it beside the original, and copy or download the words. Recognition runs in your browser.

Start with a clear image

JPEG, PNG or WebP · up to 20 MB / 20 megapixels. You can also paste an image from your clipboard onto this page.

Your screenshot or scan will appear here for comparison.

English recognition. The first run downloads about 7 MB of recognition files from PixelFix. These files may be cached; your image and results are not saved by the tool.

Review the words

Check names, numbers and punctuation carefully. OCR can make mistakes.

Choose a screenshot or scan to begin.

Useful for text trapped inside a picture

Use this for a screenshot of instructions, a photographed notice or a scanned paragraph that you need to quote or edit. Recognition produces plain text, so it will not preserve the original fonts, table borders or page design.

After extraction, compare the result with the original displayed alongside it. Correct errors in the text box before copying. The downloaded .txt file contains your edited text, encoded as UTF-8, without a watermark or extra introduction.

Get a better reading

  1. Keep the text upright. Use Rotate 90° if the source is sideways.
  2. Use a sharp original with dark text on a light background. Compressing a screenshot heavily can blur letter edges.
  3. For a cropped paragraph, try One block. Automatic works better when several text regions share a page.
  4. Check easily confused characters: O and 0, I and 1, decimal points, dates and email addresses.

This version is intended for printed English. Handwriting, decorative lettering, complex tables and other languages can produce poor or empty results. A larger image cannot restore detail that was absent from the source.

What stays on your device

The Tesseract.js recognition engine and English model download from this site when you start. Your selected image is processed locally; there is no OCR upload endpoint. Clear removes the current image and text from the interface. Refreshing loses the result, so download anything you want to keep.

Need to turn reviewed text into a document? Copy it into Text to PDF. For a PDF that already contains selectable text, use PDF Tools instead of taking a screenshot and recognizing it again.

Common questions

Why did it find no text?

The text may be too small, blurry, sideways or outside the supported language. Try a clearer source, rotate the image or select a different layout. Empty recognition is reported explicitly rather than shown as a successful conversion.

Can it read a PDF directly?

This tool accepts images. For a scanned PDF, first export the needed page with PDF to JPG, then recognize the image. Check the result against the PDF before relying on it.

Does Cancel stop the download too?

Cancel discards the current run. If the engine is still starting, its download may finish in the background; your image will not be sent for recognition by that cancelled run. You can start again with another image.