Image tool / 28
Image to Text (OCR)
Pull the text out of a photo or screenshot in your browser, in twelve languages, without uploading it.
Image to Text (OCR): TOEA runs a full optical character recognition engine compiled to WebAssembly, reads the characters in the picture, and rejoins lines into paragraphs so the result reads as prose. Runs 100% locally in your browser with zero server file uploads.
- Category
- Image tools
- Runs
- In your browser
- Cost
- Free · no sign-up
- Availability
- Ready to use
Runs entirely in your browser
PNG, JPG, WebP, or BMP · up to 30 MB, never uploaded
Choosing the language
Pick the language the text is written in before reading it. The language model supplies the alphabet and the vocabulary the engine checks its guesses against, so English selected for a French page drops or mangles accents, and a Latin-script model turns Chinese, Japanese, Korean, Arabic, or Russian text into nonsense. For a page that mixes languages, choose the main one and expect mistakes in the rest.
Checking the joined paragraphs
With lines joined into paragraphs, a hyphen at the end of a line is taken as a word split and removed, so a real compound broken there, such as well- followed by known, comes back as wellknown. Lines are joined with a space, which also puts a space at every line break in Chinese or Japanese text, where none belongs. Turn the option off for those languages and for tables, lists, addresses, and code, where the line breaks carry meaning. For a scanned PDF, use PDF OCR.
How to use it
- Choose an image.
- Pick the language of the text.
- Read the result and copy or download it.
Privacy & limitations
Your file is never uploaded. The recognition engine and language model are downloaded once — around 10 MB — and cached by your browser, then everything runs on your own machine. Hosted OCR services work the other way round: they want the document.
Related tools
Frequently asked questions
Why is the first run slow?
The engine and the language model have to download the first time, around 10 MB. After that your browser caches them and later runs start immediately.
How accurate is it?
A clean screenshot is usually near-perfect. A phone photo of a page at an angle, or small or blurry text, will contain mistakes. The confidence score above the result tells you which you are looking at, and it is always worth reading back against the original.
Can it read handwriting?
No. The engine is trained on printed type; handwriting is a different problem and it will produce nonsense.
What improves the result?
More pixels, straighter text, and higher contrast. Crop to just the text, and photograph the page flat rather than at an angle.
Free tool · runs in your browser · no account required