What is OCR Image to Text?

Extract text from images, screenshots, scanned documents, and photos using Tesseract OCR running entirely on your device. Supports multiple languages and outputs editable text you can copy or download. No files leave your device — all processing happens locally.

The tool ships with Tesseract OCR compiled to WebAssembly and 20 trained language packs (English, Spanish, French, German, Italian, Dutch, Portuguese, Polish, Swedish, Turkish, Russian, Arabic, Hindi, Thai, Vietnamese, Chinese Simplified/Traditional, Japanese, Korean, Indonesian). Each pack downloads once on first use and is cached afterwards. You can drop in an image or a single-page PDF, rotate sideways photos, and drag-select a single region before recognizing, then save the result as plain text or a Word document. Accuracy holds up well on screenshots, scans at 300 dpi, and clear receipts; expect 95%+ on print, lower on handwriting or skewed photos.

How to use

1. Upload an image (PNG, JPG, BMP, or WebP) or a single-page PDF containing the text you want to extract, or paste a screenshot from your clipboard. New here? Hit Try Sample to watch it work on an example.
2. Select the language of the text in the image for better recognition accuracy. Click 'Extract Text' to start OCR processing.
3. Review the extracted text, fix any mistakes, then copy it or download it as a .txt or Word (.docx) file.

When to use

• Pulling editable text out of a screenshot of a chat, slide, or PDF that doesn't allow copy/paste.
• Transcribing line items from a paper receipt photo into a budget spreadsheet.
• Digitising handwritten or printed lecture notes so you can search them later.

Result

Upload a photo of a restaurant receipt: the OCR pulls out each line, like 'Caesar Salad — $12.50' and 'Sparkling Water — $3.00', and gives you clean text to paste straight into a spreadsheet.

FAQ

Which languages can the OCR engine recognise?

English, Spanish, French, German, Italian, Dutch, Portuguese, Polish, Swedish, Turkish, Russian, Arabic, Hindi, Thai, Vietnamese, Chinese (Simplified and Traditional), Japanese, Korean, and Indonesian — 20 in all. Pick the closest match for your image. Mixed-language images work best when you choose the script that covers most of the text.

What kind of accuracy should I expect?

On clean print at 300 dpi or higher (screenshots, scanned PDFs), accuracy is typically 95% or better. Handwriting, low contrast, motion blur, or photos taken at an angle drop accuracy significantly. The confidence score next to the result tells you how sure Tesseract is.

Why does the language pack take a few seconds to load the first time?

Each Tesseract language model is around 10–20 MB and downloads once when you first pick that language. After the first use it stays cached in your browser, so subsequent recognitions on the same language are near-instant.

Can I paste a screenshot from the clipboard instead of uploading a file?

Yes. Take a screenshot (Win+Shift+S on Windows, Cmd+Ctrl+Shift+4 on macOS), then click Paste Image. The tool reads the image directly from the clipboard without saving a file to disk.

Will my images be sent anywhere?

No. Tesseract runs entirely inside the page via WebAssembly. The image bytes never leave your device. You can confirm this by disabling the network in DevTools and watching the recognition still complete.

Related Tools