Image to Text Converter (OCR)

Pull the text out of photos, screenshots, scanned pages and PDFs. Drop an image, paste a screenshot with Ctrl+V, and copy the result in seconds. The OCR engine runs inside your browser, so your images are never uploaded.

Drop images or a scanned PDF here

PNG, JPG, WebP, screenshots or PDF. Several files at once are fine — or paste a screenshot with Ctrl+V.

🔒 Processed on your device — nothing is uploaded.

Language of the text Text also contains English Photo cleanup Line breaks

Convert Images to Editable Text — Privately

Retyping text from a picture is slow and error-prone. This free image to text converter uses optical character recognition (OCR) to read the words in an image and turn them into plain text you can copy, edit, search or translate. It handles phone photos of documents, screenshots of error messages and chats, scanned letters, slides, posters and PDFs that are nothing but pictures of pages.

Most online OCR sites upload your file to their server. This one doesn't: the Tesseract OCR engine is downloaded to your browser once and runs on your own device. That makes it safe for contracts, IDs, payslips, medical letters and anything else you would rather not hand to a stranger — and after the first use it keeps working even when you are offline.

How to extract text from an image

Add your image — drag it onto the box, click Choose images, or copy a screenshot and press Ctrl+V (⌘+V on a Mac). You can add several images or a scanned PDF at once.
Pick the language — choose the language the text is written in. The first time you use a language its data (about 1–3 MB) is downloaded and then cached by your browser.
Copy or download — the recognised text appears next to each image. Fix anything that looks off, then copy it or save it as a .txt file.

21 OCR languages, including Indian scripts

English, Hindi, Marathi, Bengali, Tamil, Telugu, Gujarati, Kannada, Malayalam, Punjabi, Urdu, Spanish, French, German, Portuguese, Italian, Arabic, Russian, Chinese (Simplified), Japanese and Korean. Keep Text also contains English ticked for documents that mix a regional language with English words, numbers or email addresses — common on Indian forms, bills and certificates.

Tips for the most accurate OCR

Fill the frame and keep it sharp. Hold the phone parallel to the page, avoid motion blur, and crop away the table or desk around the document.
Light it evenly. Shadows and glare are the most common cause of errors. Photo cleanup flattens uneven lighting automatically; set it to Always for very dim or shadowed photos.
Small text is fine. Screenshots with tiny interface text are enlarged automatically before recognition, so you don't need to zoom in first.
Dark mode works too. Light text on a dark background is detected and inverted before OCR.
Check numbers. OCR can confuse look-alike characters such as 0/O, 1/l/I and 5/S. A low-confidence badge tells you when to double-check.

What OCR can't do (yet)

This tool reads printed and typed text. Handwriting, decorative fonts, text on curved or crumpled surfaces, and very low-resolution images give poor results. Complex layouts such as multi-column newspapers are read column by column, and tables come out as lines of text — use Image to Excel to rebuild a table as spreadsheet cells, or OCR PDF to make a whole scanned document searchable.

Private by design

Images are decoded, cleaned up and recognised entirely in your browser using WebAssembly. No file, text or thumbnail is sent to SarvKit or anyone else — you can confirm it in your browser's network panel. The only downloads are the OCR engine and language files, served from sarvkit.com and cached on your device.

Frequently Asked Questions

Yes. There is no sign-up, no watermark, no daily limit and no paid tier. Because the OCR runs on your own device there is no server cost per image, so there is nothing to limit.

No. The Tesseract OCR engine runs inside your browser using WebAssembly. Your images and the extracted text never leave your device, which makes the tool safe for confidential documents.

Take the screenshot (for example with Win+Shift+S on Windows or Cmd+Shift+4 on a Mac), click anywhere on this page and press Ctrl+V or Cmd+V. The text appears a moment later, ready to copy.

21 languages: English, Hindi, Marathi, Bengali, Tamil, Telugu, Gujarati, Kannada, Malayalam, Punjabi, Urdu, Spanish, French, German, Portuguese, Italian, Arabic, Russian, Chinese (Simplified), Japanese and Korean. Each language's data is downloaded the first time you use it and then cached.

Not reliably. The engine is trained on printed text, so neat block capitals sometimes work but joined-up handwriting usually does not. Printed and typed text gives the best results.

Yes — drop the PDF here and every page is rendered and recognised. If you want to keep the PDF but make it searchable and copyable, use OCR PDF instead; to get an editable Word document, use Scan to Word.

OCR accuracy depends on the image. Blur, glare, strong shadows, tiny text in a photo, unusual fonts or the wrong language setting all cause errors. Try a sharper photo, set Photo cleanup to Always, and make sure the selected language matches the text.

Once the page, the OCR engine and your language have loaded, recognition needs no connection at all. Your browser keeps the language data, so later visits start faster too.

Image to Text Converter (OCR)

Related Tools