Free image to text converter with OCR
This image to text tool uses optical character recognition (OCR) to pull typed or printed text out of pictures — photos of documents, scanned pages, screenshots, receipts, book pages, slides and signs. Drop an image in and you get editable text you can copy into an email, a document or a spreadsheet instead of retyping it.
It is powered by Tesseract, the open-source OCR engine originally developed at HP and later maintained with Google’s support, compiled to WebAssembly so it runs right inside your browser. Your images are never uploaded; only the language data is downloaded once and then cached on your device. It reads more than 25 languages, including Hindi, Bengali, Tamil, Arabic, Chinese and Japanese, and can combine two languages for mixed text.
How to use the Image to Text
- Choose the language of the text in “Language in the image”. For mixed text, such as Hindi with English words, also pick a second language.
- Leave “Layout” on Automatic for documents, or choose “Single line” or “Scattered text” for captions, labels and signs.
- Drop one or more images onto the box, click it to choose files, or paste a screenshot with Ctrl+V (Cmd+V on Mac).
- Follow the progress bar. The first run downloads the language data (a few megabytes), after which it is much faster.
- Check the confidence score, correct any mistakes directly in the text box, then press “Copy text” or “Download .txt”.
Image to Text features
29 languages and scripts
Latin, Devanagari, Bengali, Tamil, Telugu, Gujarati, Kannada, Malayalam, Gurmukhi, Arabic, Urdu, Cyrillic, Thai, Chinese, Japanese and Korean.
Two languages at once
Combine languages (for example Hindi + English or Arabic + English) so mixed documents are read correctly.
Batch and paste
Queue several images to be read one after another, or paste screenshots straight from the clipboard.
Automatic enhancement
Small screenshots are upscaled, colours turned to greyscale and dark-mode images inverted — simple steps that noticeably improve accuracy.
Confidence score
Each image shows Tesseract’s average confidence so you know when to double-check the result.
Private OCR
Recognition happens on your device. Nothing you scan — IDs, bills, notes — is sent to a server.
Tips for the most accurate OCR
- Use a sharp, well-lit photo. Blur and shadows are the biggest causes of errors. Hold the phone parallel to the page and tap to focus.
- Crop to the text you need. Removing tables, photos and margins with Crop Image helps the layout analysis.
- Straighten tilted photos. Text rotated by more than a few degrees is read badly; fix it first with Rotate & Flip Image.
- Aim for text at least 20 pixels tall. Tiny text in low-resolution images is hard to read — keep “Enhance image first” ticked to upscale it.
- Pick the correct language. Reading Hindi with only English selected produces gibberish. Add a second language for mixed text.
- Match the layout setting. Use “Single line” for a single caption or number and “Scattered text” for packaging, signs and posters.
What OCR can and can’t do
OCR works best on printed or on-screen text with good contrast: documents, books, invoices, menus and screenshots usually come out 95–99% correct. Handwriting, decorative fonts, text on busy photo backgrounds and very low-resolution images are much harder and can produce many errors. The result is plain text, so bold, columns and tables are not preserved — paste it into a document and reformat as needed.
For PDFs that already contain selectable text, PDF to Text is faster and exact. Use this OCR tool for scanned PDFs by first saving the pages as images with PDF to JPG.
Reading the confidence score
| Confidence | What it usually means |
|---|---|
| 80–100% | Clean result — a quick proofread is enough |
| 60–79% | Some words are likely wrong; check names and numbers carefully |
| Below 60% | Poor image, wrong language or layout — improve the photo or settings and read again |
Image to Text FAQ
How do I convert an image to text for free?
Choose the language, drop your image onto this page and wait a few seconds. The extracted text appears in the box, ready to copy or download as a .txt file. It is free with no sign-up or limits.
Is my image uploaded anywhere?
No. The OCR engine runs inside your browser using WebAssembly. Only the engine and language files are downloaded from a public CDN; your images and the recognised text never leave your device.
Can it read Hindi, Tamil or other Indian languages?
Yes. Hindi, Marathi, Bengali, Tamil, Telugu, Gujarati, Kannada, Malayalam, Punjabi and Urdu are supported. For documents that mix an Indian language with English, select both languages.
Can it extract text from handwriting?
Only neat, printed-style handwriting gives usable results. Tesseract is trained mainly on printed text, so joined-up handwriting will contain many mistakes.
Why is the extracted text wrong or empty?
The usual causes are a blurry or dark photo, the wrong language, or text that is too small. Try a sharper image, crop to the text, check the language, and switch Layout to “One block of text” or “Single line”, then press “Read again with new settings”.
How do I get text from a screenshot?
Take the screenshot, click on this page (not in a text box) and press Ctrl+V or Cmd+V. The screenshot is added and read straight away.