Image to Text
Copy text from an image with free browser OCR. Extract English or Afrikaans text from JPG, PNG and WebP, edit the result and download TXT. No image uploads.
Your images
Choose a file to begin. Processing stays on your device.
Your result
Step by step
Choose a JPG, PNG or WebP image with readable text. Select English, Afrikaans or both. Choose automatic page layout, a single text block or scattered text. Click Extract text, wait for local recognition, then proofread and edit the result. Copy it or download a plain TXT file.
What is OCR?
Optical character recognition turns the shapes of letters in an image into text characters. A screenshot contains pixels, not selectable text. OCR estimates which letters those pixels represent; it does not guarantee correct spelling or faithful formatting.
Prepare a clearer image
Use a sharp, well-lit image. Crop away unrelated content and straighten tilted text with the Photo Editor before recognition. Larger readable letters help more than an artificially enlarged blurry source. Avoid glare, heavy compression, decorative fonts and curved pages where possible.
Choose a language and layout
Use English for English text and Afrikaans for Afrikaans. Choose both only for mixed-language material. Automatic page layout is a sensible starting point; one block suits a simple paragraph, and scattered text can help with labels placed around a screenshot.
Why the first run takes longer
The page loads the local recognition engine and selected language data when you start. These files may be cached by your browser for later use. Your image is processed on your device. A slow first run does not mean the photo is being uploaded.
Proofread before using the text
Check names, dates, decimal points, account numbers and similar-looking characters such as 0/O and 1/l. Line breaks and punctuation may be wrong. The editable result lets you correct mistakes before copying or downloading. No confidence score should replace proofreading.
Can I keep tables and formatting?
The result is plain text. Columns can appear in the wrong reading order and tables may lose their structure. Crop one section at a time for complex pages. Handwritten text is not a strong use case for this version.
Limits and privacy
Use one static image up to 20 MB and 8 megapixels. English and Afrikaans are bundled; PDF input is not supported. The image and text stay in the tab. Clear removes the current workspace, and reloading loses unsaved text. Browser caches may retain downloaded engine and language files, not your recognised result.
Behind the result
OCR recognises characters rather than reproducing a document layout. Clear printed text works best. Handwriting, small letters, tables and angled photos can produce mistakes.
Decode the photo
A static image is decoded locally into a canvas. File and pixel limits bound memory use. The source image stays unchanged.
Load language data
Tesseract.js loads the bundled worker, WebAssembly engine and selected English or Afrikaans language data from this website. The browser may cache these assets.
Recognise characters
The recognition engine analyses text regions and estimates characters using its trained language data. Layout selection affects segmentation and reading order; it does not guarantee accuracy.
Review and export text
The recognised result is placed in an editable text field. Copy and TXT downloads use your edited text. Formatting, tables and handwriting can be inaccurate, so proofreading is part of the workflow.
Explore the details: Tesseract.js documentation