About the Image to Text (OCR)
This image to text converter uses OCR (optical character recognition) to read printed text in a picture and turn it into plain text. Instead of retyping a paper document or a quote from a screenshot, just drop in the image and copy the result.
It runs the open-source Tesseract OCR engine inside your browser, so your images are never uploaded. Only the first run downloads the engine and the language data (1–3 MB per language); after that they are loaded from your browser’s storage.
How to use
- Click Choose an image to pick a photo or scan, or paste a screenshot with Ctrl+V.
- Select the language(s) of the text. Choosing only the languages that actually appear makes OCR faster and more accurate.
- When the progress bar reaches 100%, the text appears in the box below. You can fix any mistakes right there.
- Click Copy text or Save .txt to take the result with you.
Tips for better OCR results
- Shoot the page straight on and well lit. Skewed or shadowed photos lose a lot of accuracy.
- Text with a cap height of 20 pixels or more reads well. For tiny print, zoom in and capture just that part (small images are upscaled automatically).
- Keep Grayscale & contrast boost on for colored backgrounds and blurry photos.
- Select only the languages in the document – extra languages can turn letters into similar-looking characters from another script.
- Tables, handwriting, vertical text and decorative fonts are less accurate.
Download size per language
| Language | Download |
|---|---|
| OCR engine (shared, once) | about 1.4 MB |
| English | about 3.0 MB |
| Spanish | about 2.1 MB |
| Japanese | about 2.0 MB |
| Chinese (Traditional) | about 1.7 MB |
| German, French, Portuguese, Italian, Korean | 0.7–1.7 MB each |
Downloaded data stays in your browser’s storage, so next time recognition starts right away. If you’re on mobile data, consider doing the first run on Wi-Fi.
FAQ
Are my images uploaded to a server?
No. Text recognition runs entirely in your browser on your own device. Only the OCR engine and language data are downloaded – your images and the extracted text are never sent anywhere.
Does it read handwriting?
Neat block letters sometimes work, but the engine is trained for printed text, so handwriting accuracy is low.
Can I extract text from a PDF?
Only image files are supported. For a scanned PDF, convert the pages to JPG with PDF to Image first. If the PDF has selectable text, copying it directly in a PDF viewer is more accurate.
Why does the first run take longer?
The first time, the OCR engine and language data have to be downloaded. After that the stored copy is used and recognition usually finishes in a few seconds; very large photos take longer.
Does it support vertical Japanese or Chinese text?
It is tuned for horizontal text, so vertical writing is recognized poorly. Horizontal sections work best.