Hindi image to text

Turn a photo, scan or screenshot in Hindi into Devanagari text you can edit and copy. This page starts with Hindi and English selected, so bilingual forms and notices are read in one go, on your own device.

Drop images here, or paste with Ctrl V

JPG, PNG, WebP, HEIC, GIF, BMP, TIFF or PDF, up to 25 MB each. Add up to 50 at once.

Read on your device. Your images aren't uploaded. How this works

No image to hand? Try a sample:

How to use it

  1. Drop in a photo, scan, PDF or screenshot, paste one with Ctrl+V (⌘V on a Mac), or try the English + Hindi sample.
  2. Keep Hindi and English selected for bilingual pages, or open the language menu (the globe) to choose Marathi or Nepali instead.
  3. Step through the underlined words to check, looking closely at matras, dots and half letters.
  4. Copy the Devanagari text, or download it as a Word document, plain text or a searchable PDF.

Typing Hindi is slow for most people, even with a good keyboard, and retyping a full page from a textbook, a government circular or a newspaper cutting can take the best part of an hour. Hindi image to text reads the page for you and gives you Devanagari text you can edit, search and paste anywhere.

It suits the kind of Hindi text people actually need to copy:

  • Textbook pages, guides and question papers for study notes
  • Circulars, notices, application forms and letters from offices and banks
  • Newspaper articles and magazine pages
  • Images forwarded on WhatsApp with a message, poem or announcement in Hindi
  • Pages from older books that were never published digitally

Why Devanagari is harder than English

In English, letters stand apart and words are separated by clear gaps, which makes them fairly easy to pick out. Devanagari works differently, and each difference gives an OCR engine something extra to get right.

  • The shirorekha. The headline along the top joins every letter in a word, so there’s no gap between letters to show where one ends and the next begins.
  • Matras in every direction. Vowel signs sit above, below, before or after the consonant. The short i (ि) is written before its consonant but typed after it, so कि has to be put back in the right order.
  • Conjuncts and half letters. क्ष, त्र, ज्ञ and श्र have shapes of their own, and half forms such as the स् in पुस्तक join onto the next letter. The reph in कर्म is just a small hook above the line.
  • Small marks that change meaning. The nukta dot (ज़, फ़, ड़), anusvara (ं) and chandrabindu (ँ) are tiny, and they’re the first details to disappear when an image is blurry or small.

The engine reads Hindi with a model trained on Devanagari, so it knows these shapes. Clear printed Hindi is usually read well. The marks above, below and around the letters are where most corrections end up.

Hindi and English on the same page

A lot of Hindi text comes mixed with English: bilingual forms, bank and railway notices, textbooks that keep technical terms in English, and signboards. This page starts with Hindi and English both selected, so lines in either script are read correctly without changing anything.

If the page is entirely in Hindi, you can untick English in the language menu (the globe), which gives the engine fewer wrong options to choose from. Marathi and Nepali, which also use Devanagari, have their own options. Each extra language makes reading a little slower, so choose only the ones on the page.

Numbers come through as they’re printed, whether that’s 1, 2, 3 or Devanagari numerals such as १, २, ३. Hindi written in English letters, as in many chats, is read as ordinary Latin text.

PDFs that copy as gibberish

Many Hindi PDFs, especially older government documents, were typed in fonts such as Kruti Dev or Chanakya. These fonts store Hindi as ordinary English keyboard characters and only look like Devanagari on screen. Copy the text and you get a string of Latin letters and symbols. Even some Unicode PDFs copy badly, with matras out of place or half letters split apart, because of how their fonts were embedded.

Image to Text App takes the text from a digital PDF page as it is, so a page like this shows the same jumble here. The fix is to read the page as a picture instead. Open the PDF, take a screenshot of the page, and paste it here with Ctrl+V. The engine then reads what you see and gives you real Unicode Hindi. Scanned PDFs need no workaround, since every page is read with OCR. See PDF to text for more on how PDF pages are handled.

Getting a cleaner result

Devanagari packs a lot of detail into a small space, so resolution matters more than it does for English.

  • Make the text big enough. Move closer to the page instead of zooming, or zoom in on the screen before you take a screenshot. Matras and dots need real pixels to survive.
  • Ask for the original image. WhatsApp compresses photos each time they’re forwarded. If a forwarded notice is blurry, the person who sent it may have a sharper copy.
  • Crop to one column. Hindi newspapers use narrow columns. Crop to one article, or one column, so lines from neighboring stories don’t get mixed together.
  • Straighten pages shot at an angle. Use four-corner straightening so the shirorekha runs level. The photo to text page has more on photographing pages.
  • Lift faded photocopies. Raise Contrast or try Black & white for old circulars and photocopied forms where the strokes have faded to gray.

Auto improve is on by default and handles dark-mode screenshots, small text, a slight tilt and uneven lighting, listing each step it took.

Checking the words it wasn’t sure about

Words the engine was unsure about are underlined, and a button jumps from one to the next. Click a line and its place lights up on the image, so you can compare quickly. In Hindi, a few mistakes come up again and again:

  • Look-alike letters such as ब and व, घ and ध, म and भ, or प and ष
  • A missing nukta, or anusvara and chandrabindu swapped
  • A half letter read as a full one, so स्त turns into सत
  • A short or long i matra (ि, ी) dropped or attached to the wrong letter

If one name or term is misread the same way throughout, find and replace fixes every copy at once. For more ways to improve accuracy, see how to get accurate OCR results.

Handwritten Hindi

The Hindi model was trained on printed text, and handwritten Devanagari varies a great deal from person to person. Handwriting mode can produce a usable draft from very neat writing with clear gaps between words, but notes written quickly usually need so many corrections that typing them is faster. The handwriting to text page explains what helps.

Where the text goes next

Copy the text into Google Docs, Word, an email or a chat, and it pastes as normal Unicode Hindi. Rich copy keeps headings and lists. You can also download:

  • Word (.docx) for editing and printing
  • Searchable PDF, which keeps the original image and adds an invisible Hindi text layer you can search and select
  • PDF, which for Hindi opens your browser’s print dialog, where you choose “Save as PDF” so every character is drawn correctly
  • Plain text or Markdown for notes apps

Private by default

Certificates, bank letters and family documents are personal. Image to Text App reads them in your browser, on your device, and the image is not uploaded. History is off unless you switch it on, and even then it stays in your browser. Read more in is online OCR private?

Questions people ask

Can it read Hindi and English in the same image?

Yes. This page starts with both selected, so a form, notice or textbook page that switches between Hindi and English is read in one pass. You can choose up to four languages at once from the language menu.

Does it work for Marathi, Nepali or Sanskrit?

Marathi and Nepali have their own options in the language menu, so choose the one that matches the page. Sanskrit isn't in the list. Hindi is the closest option, but expect more corrections in the dense conjuncts Sanskrit uses.

Why does text copied from my Hindi PDF look like random English letters?

The PDF was probably typed in an older font such as Kruti Dev, which stores Hindi as Latin characters. Image to Text App takes a PDF's text as it is, so you'd get the same jumble. Take a screenshot of the page and paste it here instead, and it is read as an image into proper Unicode Hindi.

Can it read handwritten Hindi?

Only to a limited extent. The Hindi model was trained on printed text. Neat, separated handwriting in Handwriting mode can give a rough draft, but most handwritten Hindi needs a lot of correcting.

Do I need a Hindi font or keyboard?

Not to read the image. The result is standard Unicode Devanagari that displays in any modern app. To type corrections yourself, turn on a Hindi keyboard in your phone or computer's settings.

Is my document uploaded?

No. The image is read in your browser, on your device. The Hindi language file is downloaded once, the first time you read Hindi, and then kept by your browser.