Image to Text
Extract printed text from a single image with a local OCR engine. Your image stays in this browser; choose its main language for recognition.
A QUICK WALKTHROUGH
How to use this tool
- Choose one JPEG, PNG, or WebP image up to 3 MB and 4 megapixels.
- Select the main language visible in the image, then start recognition. The OCR engine and language data load from this site when needed.
- Review and edit the recognized text, then copy it or download a UTF-8 text file.
Local OCR with four language models
Recognition runs on your device using Tesseract.js and language data served by this site. The image is not uploaded or sent to an AI service. English, Simplified Chinese, Japanese, and Korean are available; choose the main language in the image.
Bounded image and text sizes
One JPEG, PNG, or WebP image up to 3 MB and 4 megapixels is accepted. The page refuses output over 120,000 characters rather than silently truncating it. The OCR engine and the selected compressed language model are fetched from this site on first use and may be cached in your browser.
OCR is not exact transcription
Recognition quality depends on focus, contrast, orientation, font, handwriting, and language. This version targets readable printed text, does not translate, preserve layout, identify document fields, or guarantee accuracy. Review every result before relying on it.
GOOD TO KNOW
Common questions
Is my image uploaded?
No. The image is decoded and processed in a worker on your device. Only the site’s OCR code and selected language model are downloaded from tools.hellokeyboard.com.
Which languages can I recognize?
Choose English, Simplified Chinese, Japanese, or Korean. Select the language that accounts for most of the visible text; the tool does not translate.
Why is the first run slower?
Your browser needs to load the OCR engine and selected language model from this site. Tesseract.js may cache them in this browser for later use.
Can it read handwriting or preserve the page layout?
It is intended for printed text and returns editable plain text. Handwriting, columns, tables, rotation, and unusual fonts can reduce accuracy or reading order.