Image to Text (OCR)
Extract text from a photo, screenshot or scan — OCR on your device in 17 languages.
About this tool
Optical character recognition without the upload: the open-source Tesseract engine runs inside your browser, reads printed text from a screenshot, scan or photo, and hands you plain text to copy or save. Seventeen languages. Free forever, no account.
How it works, and what it's honest about
The first press downloads the engine (~5 MB) and your language's data (~4 MB) from the jsDelivr CDN — code and dictionaries, never your image — after which everything runs locally and offline. Accuracy tracks the input: a straight, sharp, high-contrast picture of printed text reads near-perfectly; small type, blur, glare and photos taken at an angle read worse; handwriting is hit-and-miss, because Tesseract was trained on print. The practical fixes are all on your side of the lens — fill the frame, hold the phone square to the page, avoid shadows. Images are upscaled toward a 300-DPI equivalent before reading, which helps small text. For scanned PDFs, the PDF OCR tool runs this page by page; for PDFs that already contain real text, the extractor is instant. Count or tidy the result with the word counter and the text cleaner.
Frequently asked questions
Is my image uploaded for OCR?
No. The engine and language data are downloaded to your device once (from a CDN) and the recognition runs in your browser. The image never leaves your machine — the page works offline after the first use.
Which languages are supported?
Seventeen: English, Spanish, French, German, Portuguese, Italian, Dutch, Polish, Turkish, Indonesian, Vietnamese, Russian, Japanese, simplified Chinese, Korean, Arabic and Hindi. Each language's data downloads on first use.
Does it read handwriting?
Honestly, rarely well. Tesseract is trained on printed text; neat block capitals sometimes work, cursive mostly doesn't. Printed documents, screenshots and signs are its strength.
Why is the text garbled?
Usually image quality or the wrong language selected. Try a sharper, straighter, higher-contrast picture — and check the language dropdown matches the text.