Extract text from screenshots, photos and scans in 33 languages, including Urdu and Arabic, then edit, copy or download it as a .txt file.
Images never leave your device. Only the OCR engine and language files are downloaded from a public CDN the first time.
Drop images or screenshots here
JPG, PNG, WebP, GIF or BMP, plus HEIC and TIFF in Safari. Up to 50 images, 50 MB each.
or paste a screenshot with Ctrl+V (⌘V on a Mac)
Settings changed since these images were read.
Some images could not be read.
What the image to text converter does
An image to text converter reads the words in a picture and turns them into text you can edit, copy and search. This one uses optical character recognition (OCR) to extract text from screenshots, photos, scans and JPG or PNG files in 33 languages, ready to copy or download as a .txt file. It runs in your browser, so your images are never uploaded. It reads text only; it does not describe what a picture shows.
- Input: JPG, PNG, WebP, GIF (first frame) and BMP, plus HEIC and TIFF in Safari. Up to 50 images at a time, 50 MB each, or a pasted screenshot.
- Output: editable text for each image, with a confidence score and a box around every word found. Copy it, or download a .txt file or a ZIP.
- Languages: 33, including English, Urdu, Arabic, Hindi, Bengali, Chinese and Japanese, with up to 3 at once for mixed text.
- Limits: printed and on-screen text reads best. Handwriting, decorative fonts and tiny or blurred text are read poorly. PDF files are not read: take a screenshot of the page instead.
- Privacy: images never leave your device. Only the OCR engine and language files are downloaded, the first time.
How to convert an image to text, step by step
- Under Settings, add any language in your image, such as Urdu or Arabic. English is already selected.
- Drop your images on the box above, click Choose images (on a phone, tap it to pick a photo or screenshot), or paste a screenshot with Ctrl+V.
- Wait while each image is read. The first use downloads the engine and language data: about 4.4 MB for English, with the size for your languages shown under Settings.
- Check the text against the preview. Highlight unsure words shows the words the engine doubted; correct them in the text box.
- Copy the text, or download it as a .txt file. With several images, download one combined file or a ZIP.
How to copy text from a screenshot
No need to save a screenshot first: copy it to the clipboard with one of these shortcuts, click on this page and press Ctrl+V (⌘V on a Mac).
| Device | Shortcut that copies a screenshot to the clipboard |
|---|---|
| Windows 10 and 11 | Windows logo key + Shift + S, then drag over the area you want |
| Mac, part of the screen | Control + Shift + Command + 4, then drag over the area |
| Mac, whole screen | Control + Shift + Command + 3 |
Tips to extract text from an image accurately
| Problem | What to try |
|---|---|
| Small or blurry text | Keep Enlarge small text on, and scan paper at 300 dpi or more, which the engine reads best. |
| Light text on a dark background, such as a dark mode screenshot | Keep Fix light text on dark backgrounds on. |
| Faded scan or low-contrast photo | Keep Greyscale and boost contrast on, and take photos in even light without shadows. |
| Text is sideways or upside down | Click Rotate until the text is upright. |
| Two languages in one image | Add each language. One you add goes before English, which stays for English words. |
| Labels scattered over a picture | Set Layout to Scattered text. |
| One line, such as a code or serial number | Set Layout to A single line. |
The confidence score is the engine’s own estimate, averaged over the words it read, with longer words counting for more. It turns green at 85% or more, amber from 65% and red below; words under 65% are marked as unsure. A low score usually means blur, slanted text or a missing language.
Urdu OCR and other languages
Urdu, Arabic, Persian and Hebrew are written right to left and appear that way automatically. When Urdu and English are mixed, each line keeps its own direction.
Urdu is usually written in the Nastaliq style, where each word slopes down from right to left. Nastaliq is much harder for OCR than the plainer Naskh style. In a 2025 study of printed Urdu books, every engine tested made more mistakes on Nastaliq, and Tesseract, the engine used here, lost the most accuracy. Expect to correct Urdu text, especially Nastaliq; clear, large print works best.
| Language | Direction | First download |
|---|---|---|
| English | Left to right | 3.0 MB |
| Urdu | Right to left | 1.0 MB |
| Arabic | Right to left | 1.7 MB |
| Hindi | Left to right | 1.4 MB |
| Bengali | Left to right | 1.4 MB |
| Chinese (Simplified) | Left to right | 1.7 MB |
Sizes are for the language data; the engine adds about 1.4 MB on first use. The full list: Arabic, Bengali, Chinese (Simplified and Traditional), Dutch, English, Filipino, French, German, Greek, Gujarati, Hebrew, Hindi, Indonesian, Italian, Japanese, Korean, Malay, Marathi, Persian, Polish, Portuguese, Punjabi, Russian, Spanish, Swedish, Tamil, Telugu, Thai, Turkish, Ukrainian, Urdu and Vietnamese.
How it works and your privacy
Text is recognised by Tesseract, an open-source OCR engine first developed at Hewlett-Packard and later by Google, through tesseract.js, which runs it in the browser with WebAssembly. Your browser downloads the engine on first use, and each language’s data the first time you pick it, from cdn.jsdelivr.net, a public CDN. Both are then cached, so with the page open it keeps working if your connection drops. No image or text is sent anywhere.
Frequently asked questions
Is this image to text converter free?
Yes, with no sign-up and no daily limit. You can read up to 50 images at a time, each up to 50 MB.
Are my images uploaded to a server?
No. The text is read in your browser, on your device, and no image or text is sent anywhere.
What is OCR?
OCR, or optical character recognition, is software that finds letters in an image and turns them into text a computer can edit and search. This tool uses Tesseract, an open-source engine that recognises more than 100 languages.
Can it read handwriting?
Not well. Tesseract is designed for printed text, and its own documentation says handwriting won’t work very well, so expect to correct most of it.
How do I convert a JPG to text?
Drop the JPG on the box at the top of this page or click Choose images. The text appears with a preview of the image, ready to copy or download as a .txt file.
How do I get the text into Word?
Paste the copied text into Word or Google Docs, or open the downloaded .txt file in Word. Choose Join lines first if you want flowing paragraphs. Bold, tables and columns are not kept.
Can I extract text from a PDF?
Not directly: this tool reads images, not PDF files. Take a screenshot of the PDF page with one of the shortcuts above and paste it here. If you can already select the text in the PDF, copy it from there instead.
Why is some of the text wrong?
Mistakes usually come from blur, small text, unusual fonts, slanted text or a language that is not selected. Highlight unsure words shows likely mistakes; fix them, or improve the image and read it again.
Built and maintained by the GradientCo team. Last updated 7 October 2026. Found a problem? Tell us.