How to Convert a Scanned Hindi PDF to Word

A scanned Hindi PDF is really a photograph of a page, so Word cannot pull text out of it directly. Run it through OCR first to turn the image into selectable Devanagari text, then move that text into Word. Skipping the OCR step and converting the scan straight to Word usually gives you a blank page or a garbled result.

Why a direct PDF to Word conversion fails on scans

Regular PDF to Word tools look for an existing text layer inside the PDF and rebuild the document around it. A scanned page has no text layer at all, only a picture of the page, so there is nothing for the converter to extract. This applies to any scanned document, but it is more obvious with Hindi, since a failed conversion is more likely to show empty boxes or unreadable characters than with English text.

Step by step

1. Open the OCR PDF tool and add your scanned Hindi PDF.
2. Select Hindi as the OCR language if the tool asks, so it recognises Devanagari characters rather than guessing at Latin ones.
3. Run the OCR and review the recognised text against the original scan, checking a few lines on each page for accuracy.
4. Export the result, or open the PDF to Word converter on the newly OCR'd file to get a DOCX you can edit directly.
5. Open the DOCX in Word or Google Docs and set the font to a Unicode Devanagari font before editing further.

Try it here

OCR PDF, set to Hindi

Document language is set to Hindi + English. Then open the result in PDF to Word.

Files are uploaded over HTTPS, processed and deleted from our server straight after. Open the full OCR PDF page

Single images instead of a full PDF

If you only have a photo or a screenshot of a Hindi page rather than a full scanned PDF, the Hindi image to text tool is a faster route, since it is built specifically for Hindi OCR from a single image and also reads any English words mixed into the page.

Getting matras right

Devanagari text relies heavily on matras, the vowel signs attached to consonants, and these are exactly what a low quality scan tends to lose or blur. A matra that is even slightly faded or cut off in the scan can get recognised as a different character or dropped entirely, which changes the meaning of the word.

• Scan or photograph the original page at a higher resolution if you have access to it, rather than relying on an old low-quality copy.
• Photograph in good, even light and hold the camera flat and parallel to the page to avoid distortion at the edges.
• After OCR, read through the text rather than trusting it blindly, paying particular attention to words with compound conjuncts, which are the most likely to be misread.

Fonts that actually display Hindi correctly

Older Hindi typing systems used non-Unicode fonts, such as Kruti Dev or DevLys, where each key produces a certain shape but the underlying character code does not match standard Unicode Devanagari. Text produced this way looks correct on a computer with that exact font installed, but turns into gibberish everywhere else, including inside Word if a different font is applied, or when the text is pasted into a website. OCR output, on the other hand, should already be standard Unicode Devanagari, so use any modern Unicode Hindi font, such as Mangal, Nirmala UI or Noto Sans Devanagari, when formatting the Word document, and avoid re-typing over the OCR'd text using a legacy font.

Handwritten Hindi versus printed Hindi

OCR is generally reliable on printed Hindi text, such as a photocopied notice, a textbook page or an official notification. Handwritten Hindi is a much harder problem for any OCR engine, since handwriting varies enormously between people. If your source document is handwritten, expect a noticeably higher error rate, and budget more time for proofreading, or consider retyping short handwritten notes manually rather than relying on OCR for accuracy-critical content.

Keeping the original formatting

OCR focuses on recognising characters, not necessarily preserving the exact layout of tables, columns or letterheads from the original scan. For a simple paragraph-style document, this rarely matters. For a form, a table-heavy notice or a multi-column layout, expect to spend some time in Word afterwards rearranging text boxes, tables or spacing to match the original, since OCR output tends to flatten complex layouts into a simpler, linear structure.

When to just retype instead

For a short document, such as a one page notice or a certificate with only a few lines of Hindi text, it is sometimes faster to simply read the scan and retype the content directly in Word using a Unicode font, rather than running the full OCR and cleanup process. OCR earns its place on longer documents, multi-page notices or anything you would rather not retype by hand, where the time saved on typing outweighs the time spent proofreading the OCR output.

Common mistakes

• Converting a scanned PDF straight to Word without OCR, and getting an empty or unreadable document.
• Not selecting Hindi as the OCR language, so the tool tries to match Devanagari shapes against a Latin alphabet and produces nonsense.
• Trusting the OCR output completely without a quick read-through, especially for names, numbers and legal or exam-related documents where a wrong character actually matters.
• Applying an old non-Unicode typing font to properly OCR'd Unicode text, which breaks the display.

Tools for this

Frequently asked questions

Sources

• Google Fonts, Noto Sans Devanagari (checked 2026-09)

Rules and limits change. Always check the latest official notification or portal before you submit a form.