image to text (ocr)

read the text in a photo, screenshot or scanned page and copy it out.

runs on your device · free · no sign-up

loading tool…

how to image to text (ocr)

  1. choose the language of the text.
  2. drop a photo, screenshot or scanned pdf on the box.
  3. wait while the engine loads and reads the text.
  4. edit anything that needs fixing, then copy it or download a txt file.

optical character recognition looks at the shapes of letters and turns them into text you can search, copy and edit. this tool uses tesseract, an open-source engine, running in your browser through webassembly. the language data is about 10 mb on the first use and then stays cached.

quality of the input matters most. photograph pages straight on in good light, or take a screenshot instead of a photo of a screen. text should be at least about 20 pixels tall. dark text on a light, plain background works best. handwriting is not reliable with this engine.

features

  • over twenty languages
  • reads images and scanned pdfs
  • edit the result, copy it or download a txt file
  • 100% in your browser: no upload, no account, no watermark

popular uses

frequently asked questions

how accurate is it?

clear, high-contrast printed text works best. handwriting, blur and small fonts reduce accuracy.

which languages are supported?

english by default, and many more through the language menu, including spanish, french, german, hindi, arabic, chinese, japanese and korean. the language data downloads on first use.

is my image uploaded?

no. recognition runs in your browser; only the open-source engine files are downloaded.

how can i get a better result?

use the sharpest, largest image you have, with the text straight and evenly lit. crop away anything that is not text, and pick the right language before you start.

can it read a pdf?

yes, you can drop a scanned pdf and the first pages are read. image files work as well.

why is the text full of mistakes?

usually the picture is too small, blurry or skewed, or the wrong language is selected. photos of curved pages and decorative fonts are the hardest.

related tools