Image to Text Converter (OCR)

Extract text from screenshots, photos and scanned documents in 13 languages. Paste or drop an image, then copy or download the text — your image stays in your browser.

Drop an image with text here

or click to browse — PNG, JPG, WebP · you can also paste a screenshot (Ctrl+V / ⌘V)

🔒 Your image is processed in your browser and never uploaded. Only the open-source OCR engine and language files are downloaded from a public CDN.

Frequently Asked Questions

How do I extract text from an image for free?

Drop a PNG, JPG or WebP image into the tool or paste a screenshot with Ctrl+V (⌘V on Mac), choose the language of the text and click Extract text. The recognized text appears in an editable box that you can copy or download as a .txt file.

Is my image uploaded to a server?

No. Recognition runs inside your browser with Tesseract compiled to WebAssembly, so the image never leaves your device. The OCR engine and the language data (about 2–15 MB depending on the language) are downloaded once from the public jsDelivr CDN and then cached by your browser.

Which languages are supported?

English, Spanish, French, German, Italian, Portuguese, Dutch, Polish, Turkish, Russian, Simplified Chinese, Japanese and Arabic. Pick the language the text is written in: using the wrong one noticeably lowers accuracy, especially for accented letters and non-Latin scripts.

Can it read handwriting?

Only neat, printed-style handwriting gives usable results. Tesseract is trained on printed fonts, so cursive or messy handwriting will produce many errors.

What does the confidence percentage mean?

It is Tesseract's average certainty across all recognized words. Above 85% the text is usually near-perfect; below 60% you should proofread carefully or improve the image by cropping, enhancing or using a higher resolution.

How can I improve OCR accuracy?

Use a sharp, well-lit image where capital letters are at least 20 pixels tall, crop away everything that is not text, keep lines horizontal and turn on Enhance image for low-contrast photos or small screenshots.

How OCR Works

Optical character recognition turns a picture of text into real, editable characters. This tool uses Tesseract, the open-source engine originally developed at HP and later maintained by Google, compiled to WebAssembly so it runs in your browser. Recognition happens in a few stages:

Binarization — the image is converted to black text on a white background.
Layout analysis — the engine finds blocks, columns, lines and words.
Line recognition — an LSTM neural network reads each text line as a sequence, using the trained model for the selected language.
Post-processing — words are checked against the language's dictionary and each gets a confidence score.

Because the neural network is trained per language, choosing the correct language matters. An English model does not know the Spanish ñ or German ß, and it cannot read Cyrillic, Arabic or Chinese at all.

Tips for Better OCR Accuracy

Resolution matters most. Tesseract works best when capital letters are 20–40 px tall. Photograph documents close-up, or scan at 300 DPI.
Crop to the text. Logos, photos and table borders confuse layout analysis. Crop first, then run OCR.
Keep it straight. Lines tilted more than a few degrees lose accuracy quickly; rotate the photo so text is horizontal.
Use Enhance for poor contrast. Grey text on a coloured background, shadows on paper and small screenshots benefit from the grayscale + contrast + upscale option.
Avoid heavy JPEG compression. Artifacts around letters are read as noise; PNG screenshots give the cleanest results.

Common Uses

Copy an error message from a screenshot a colleague sent you, grab a quote from a photographed book page, digitize a printed receipt or invoice, extract text from a slide in a meeting recording, or pull a Wi-Fi password off a photo of a router label. Since nothing is uploaded, it is also suitable for documents you would rather not send to an online service, such as contracts or medical letters.

What Is Downloaded, and What Is Not

To be transparent: the first time you run OCR, your browser downloads the Tesseract engine and the trained data for the selected language from jsDelivr, a public CDN for open-source packages. Together that is roughly 2–15 MB, depending on the language (models for scripts with thousands of characters are larger). The files are cached, so later runs start much faster. Your image and the recognized text are never sent anywhere.

Limitations to Keep in Mind

OCR reads characters, not meaning. Complex tables lose their column structure and come out as plain lines of text; multi-column magazine layouts may be read in the wrong order; mathematical formulas, stylised logo fonts and text over busy photos are often misread. Similar-looking characters such as O and 0, l and 1, or rn and m are the most common mistakes, so proofread numbers, codes and passwords before using them. For scanned PDFs, convert the pages to images first and run OCR on each one.

Crop Image — isolate the text area before running OCR.
HEIC to JPG — convert iPhone photos so any browser can read them.
Image Resizer — upscale tiny screenshots for better recognition.
Merge Images — stitch several screenshots into one picture.
JPG to PDF — keep the original scans together in one document.