Image to Text

Add a photo, a screenshot or a scanned page to the box above, pick the language of the text and press Extract text. A moment later the words appear in a text box, ready to correct, copy or save as a Word or TXT file.

The reading is done by your own browser. Your pictures are not uploaded anywhere. The only thing downloaded is the language data the reader needs, once for each language you choose.

Many tools that read text from pictures let you try one page and then ask you to pay or sign up. This one is an image to text converter free for everyone, with no account, no daily limit and no watermark on the files you save.

It stays free because your own device does the work. There is no server reading your pictures, so there is no cost per page to pass on to you. Add up to 30 files at a time, as often as you like.

Free does not mean basic. You get 24 languages, a clean-up step that makes faint pictures easier to read, a choice of layouts for different kinds of text, and a text box where you can fix mistakes before you copy or save.

When you extract text from image online, most websites send your picture to their server, read it there and send the text back. That is slow on a weak connection, and your picture sits on a computer you do not control.

Here the text reader runs inside the browser tab, using WebAssembly. Your picture is read on your phone or computer and never leaves it. That matters for ID cards, bank letters, medical reports and anything else private.

It works the same on Windows, Mac, Linux, Chromebook, iPhone and Android, in any modern browser. Nothing to install, nothing to update.

Because nothing is sent to a server, there is also no waiting in a queue. The speed depends on your device and the size of the picture. A single page usually takes a few seconds on a phone and less on a computer.

The reader knows 24 languages. Fourteen of them are languages of India and nearby: English, Hindi, Marathi, Bengali, Gujarati, Punjabi, Tamil, Telugu, Kannada, Malayalam, Odia, Urdu, Nepali and Sanskrit. The other ten are Arabic, Spanish, French, German, Portuguese, Italian, Russian, Chinese (Simplified), Japanese and Korean.

You can choose up to three languages at once. Pick every language that appears in the picture. A form with English labels and Hindi answers reads best with both English and Hindi selected. A page in one language reads best with only that language, because extra languages give the reader more chances to guess wrong.

Each language is downloaded the first time you use it. The files are between 2 and 15 MB per language, so the first read can take a little longer on a slow connection. After that your browser usually keeps them, and the next read starts quickly.

Hindi image to text works with printed Devanagari: newspapers, books, government letters, notices, forms and screenshots from phones. Select Hindi in the language list. If the page also has English words, select English too.

If your browser is set to Hindi, the tool starts with English and Hindi already selected. Matras, half letters and conjuncts are read well when the print is clear and the picture is sharp. Small or blurry text loses matras first, so take the photo close and in good light.

The same goes for Marathi, Bengali, Gujarati, Punjabi, Tamil, Telugu, Kannada, Malayalam, Odia, Urdu, Nepali and Sanskrit. When you save the result, the TXT file is written as UTF-8 so the letters show correctly in Notepad, WhatsApp and Word. For a Hindi PDF with real text inside, Hindi PDF to Word keeps more of the layout.

Before reading, the tool can tidy the picture. Auto, the recommended setting, turns it grey, stretches the contrast so dark letters stand out from the background, and enlarges small pictures so the letters are big enough to read. This alone fixes most dull phone photos.

Black & white goes further. It turns every point of the picture into pure black or pure white. That helps with faint photocopies, yellowed paper, pencil-grey print and photos of paper under room light. On a busy or coloured background it can remove too much, so try Auto first.

As it is sends the picture to the reader without changes. Use it when the picture is already clean, such as a sharp screenshot, or when Auto makes coloured text vanish.

Text comes in different shapes, so the reader has four layouts. Auto finds columns, headings and paragraphs on its own and suits most pages. One block reads everything as a single block of text, which helps when Auto splits a neat paragraph in the wrong places.

One line is for a single line of text, such as a serial number, a number plate or a heading cut out of a picture. Scattered text is for words spread around the picture: screenshots of apps, signs, labels, receipts and posters.

Line breaks can be kept as in the picture, or joined into paragraphs. Keep suits lists, addresses, poems and receipts. Join suits book pages and articles: lines inside a paragraph are joined, blank lines stay as paragraph breaks, and words split with a hyphen at the end of a line are put back together.

JPG to text is the most common job, because most phone photos and scans are JPG files. Add the JPG and the text comes out in seconds. PNG, WebP, GIF, BMP, AVIF and JFIF pictures work the same way.

HEIC photos from an iPhone are read directly, with no need to convert them first. If you also want a JPG copy of your iPhone photos for other uses, HEIC to JPG does that.

Each file can be up to 50 MB. Very large photos are scaled to a size the reader handles well, and small ones are enlarged when the clean-up is on Auto.

Screenshot to text is the quickest way to copy words from places that do not let you select them: a video, an app, a photo of a slide, a PDF that is only a picture, or an error message.

Take the screenshot, click on this page and press Ctrl+V, or Cmd+V on a Mac. The screenshot is added at once, with no need to save it as a file first. Then press Extract text.

Screenshots are usually sharp, so As it is or Auto both work. If the words are spread around the screen, choose Scattered text in the layout options.

Photo to text on iPhone works right here in Safari or Chrome, with no app to install. Tap Select images and choose a photo from your library. On most phones the picker can also open the camera so you can take the photo there and then.

iPhone photos are often HEIC files, and they are read as they are. The result appears in a text box you can edit with the phone keyboard. Tap Copy text and paste it into Notes, WhatsApp, Mail or anything else, or save it as a Word or TXT file.

Android phones work the same way. For the best result, hold the phone straight above the page, fill the screen with the text, avoid shadows and tap the screen to focus before you take the photo.

Please read this before you try handwriting to text. The reader in this tool is trained on printed text. It reads typed and printed letters well, but its handwriting recognition is limited.

Neat, clear writing can work: separate block capitals, large letters, dark ink on plain white paper, one line at a time. Joined-up cursive writing, quick notes, doctor's prescriptions and small or slanted writing usually come out as wrong letters or nonsense.

If you want to try, write or photograph with a dark pen on white paper, choose Black & white in the clean-up options and One block or One line in the layout. Then check every word in the text box. For long handwritten pages, typing them yourself is often faster than fixing a poor result.

Scanned image to text is what OCR was first made for. A scanner or a scanning app gives you a picture of a page. The reader turns that picture back into words you can search, copy and edit.

You can add scanned PDFs as well as pictures. The tool reads up to 50 pages of each PDF, page by page, and marks where each page starts in the result. Password-protected PDFs need to be unlocked first.

If your PDF was made on a computer and already contains real text, OCR is not needed. PDF to Text copies that text directly, and PDF to Word keeps the layout too. Use this tool when the PDF is only pictures of pages.

For image to text copy paste, press Copy text under the result. The whole text goes to your clipboard, and you can paste it into an email, a chat, a document or a form.

Before you copy, you can correct the text right in the box. Fix a wrong letter, delete a stray line or add a missing word. The copy and the saved files always use what is in the box, with your changes.

If you added several pictures, choose a file in the list above the text to see and copy its text.

To get image to text in Word format, press the Word button under the result. You get a .docx file that opens in Microsoft Word, Google Docs, LibreOffice and WPS Office. Each line of the text becomes a paragraph, on an A4 page.

The Word file holds the words, not the look of the original. Fonts, colours, tables and pictures are not rebuilt. That keeps the file clean and easy to edit, but if you need the layout of a PDF with real text inside, PDF to Word is the better choice.

The TXT button saves plain text instead, which opens anywhere. By default the files are named after your picture with -text added, and you can turn that off in the options.

Add up to 30 files at once: pictures, scanned PDFs or both. They are read one after another, with a progress bar on each thumbnail. You can stop at any time, and the files already read keep their text.

When there is more than one file, two extra buttons appear: All in one TXT and All in one Word. They join the text of every file into one document, with the file name above each part.

Before you start, you can turn any picture that is on its side with the small turn button on its thumbnail. Each tap turns it a quarter turn. An upright picture reads far better than a sideways one.

You can also remove a file you added by mistake with the small cross on its thumbnail, or press Clear all to start again. Ctrl+Enter starts the reading from the keyboard.

Under the result you see the number of words and an accuracy score. The score is how sure the reader is about what it read, not a promise. A clean printed page usually scores high. A dark, blurry or tilted photo scores lower.

When the score is below 60%, the tool suggests what to try: Black & white, turning the picture upright, or checking the language. Those three fixes solve most poor results.

Always read the text through before you use it for something important. Numbers, names and dates are the places where one wrong character matters most.

OCR reads a picture much like you do. If you have to squint to read a word, the reader will struggle with it too. The clearer the picture, the fewer mistakes you need to fix.

The best pictures have dark text on a light, plain background, letters that are not too small, even light across the page and the page lying flat. A photo taken straight from above is far better than one taken at an angle, because tilted lines and shrinking letters confuse the reader.

Text on top of a photo, glossy pages with glare, curled book pages near the spine and very fancy fonts are harder. Try to get a flat, evenly lit shot, and crop away the parts you do not need before you start.

Size matters too. For printed text, letters that are at least about 20 pixels tall read well. Auto enlarges small pictures for you, but it cannot add detail that was never there, so a sharp photo always beats an enlarged blurry one.

Your pictures are not uploaded. They are opened and read inside your browser, and the text stays on your device until you copy or save it. Close the tab and everything is gone.

The only download is the reader itself and the language data, which comes the first time you use a language. Nothing about your pictures is sent with it. That makes the tool safe for ID cards, certificates, bank papers and private letters.

Your settings, such as the languages you picked, are kept in your own browser so the next visit starts the same way. They are not linked to you and you can clear them with your browser data.

The tool uses Tesseract, a well-known open-source OCR engine, running in your browser as WebAssembly. OCR stands for optical character recognition: software that looks at the shapes in a picture and works out which letters they are.

First the picture is turned upright and, on Auto, made grey with stronger contrast and enlarged if small. Scanned PDF pages are drawn at print quality first. Then Tesseract reads the picture with the languages and layout you chose, and the text is tidied: extra spaces at line ends are removed and, if you asked, lines are joined into paragraphs.

Take the photo in daylight or good room light, with no shadow of your hand or phone across the page.

Hold the phone straight above the page, not at an angle, and fill the screen with the text you need.

Crop away anything that is not text, such as the table, your fingers or a busy background.

Choose only the languages that really appear in the picture.

Use Black & white for faint scans and old photocopies, and As it is for clean screenshots.

Read the result through and fix mistakes in the text box before you copy or save.