Word to PDF in Hindi and other Indian languages with every letter in place

The short answer: nine scripts work, if the text is Unicode

Yes, a Word file in Hindi, Bengali, Gujarati, Punjabi, Odia, Tamil, Telugu, Kannada or Malayalam can become a PDF with its letters joined and its vowel signs in the right places. You need two things: a .docx file, and text typed as Unicode, which is what phone keyboards type. Then:

1. Open Word to PDF in your phone's or computer's browser.
2. Tap Choose Word files and pick your .docx document.
3. Wait a little. The first Indian-language document also downloads a text engine and the fonts for its script, so it takes a moment longer than an English one.
4. Press Download. The PDF gets your document's name.
5. Zoom in and check the joined letters, as described in the check list.

Your document is converted on your own device and never uploaded. There is no sign-up and no watermark. If your text comes out as English letters instead of Hindi, the cause is an old font, and a later part shows how to test for it and fix it.

If you have Word on a computer, its own Save As PDF is made by the same program that laid out the document, so use that when the exact layout matters. This page is for the times Word isn't at hand: on a phone, on someone else's computer, or when a file arrives on WhatsApp.

Which languages work, and which don't

The converter works by script, the style of writing, not by language. Several languages share one script, so nine scripts cover many languages:

Each test notice had a bold heading, justified paragraphs, English words in brackets, the script's own digits or ordinary ones, a table and a page number at the bottom. We copied the text back out of every PDF with two different PDF readers, Poppler and PDFium (the one inside Chrome). Every paragraph and every table cell came out word for word.

What doesn't work yet

Urdu and other right-to-left text, Sinhala, Chinese, Japanese, Korean, Thai and Tibetan aren't drawn yet. Text in those scripts shows as empty boxes, and the row says what kind of text it couldn't draw. For an Urdu document, use a program that supports right-to-left writing, such as Word or LibreOffice, to save the PDF.

Why Indian-language text breaks in some PDFs

In English, one key gives one letter, and each letter sits beside the one before. Indian scripts don't work that way. Letters join into conjuncts, like क्ष, त्र and ज्ञ in Hindi. A vowel sign can sit before, above or below the letter it belongs to. In हिंदी, the sign ि is typed after ह but drawn before it. What you see is not always in the order it was typed.

To draw this right, a program needs two things: a shaping engine, which works out which shapes go where, and a font that has those shapes. When a converter lacks one of them, the PDF goes wrong in one of three ways:

• Empty boxes. The font has no shape for the letters.
• Letters side by side that should have joined, and vowel signs in the wrong place. The shaping step is missing.
• English letters where Hindi should be. The document was typed in an old font. This is a problem with the text itself, and the next part deals with it.

Our converter uses HarfBuzz as its shaping engine. It is the text engine used by Chrome, Firefox, Android and LibreOffice. It draws each script with Google's Noto Sans fonts and puts the letters it uses inside the PDF. That is why the PDF looks the same on a phone that has no Indian fonts of its own. Word does this job with Windows fonts such as Mangal, Latha and Nirmala UI. Google Docs and LibreOffice also support Indian scripts, and the article on converting without Office compares them.

The most important check: is your text Unicode?

Unicode is the standard that gives every Hindi letter, every Tamil letter and so on its own code. Phones, Google's Indic keyboards and the Indian keyboards in Windows all type Unicode, so most documents are.

Older documents, such as many office files, were typed in fonts like Kruti Dev, DevLys, Chanakya or Bamini. Those fonts are not Unicode. They keep English letters that only look Indian in that one font. Take the font away and what is left is English gibberish, and in a PDF that gibberish stays.

How to tell, in a minute

1. Check the font name. Open the document in any program and click on a line. Names like Kruti Dev 010, DevLys 010, Chanakya or Bamini mean an old font. Names like Mangal, Nirmala UI, Latha, Vrinda or Gautami are Unicode fonts.
2. Copy a line and paste it into a chat or a search box on your phone. Unicode text pastes as Hindi. Old-font text pastes as English letters, such as "vkidk" where आपका should be.
3. Think about how it was typed. A phone keyboard or the Hindi keyboard in Windows gives Unicode. Text typed on an English keyboard with Kruti Dev or DevLys chosen as the font, the way many offices and typing classes worked, is old-font text.

Our converter also looks at the font names. When a document uses an old Indian font, the row says so and explains that the text will come out as English letters.

How to fix it

• Use a "Kruti Dev to Unicode" converter (search for it by that name), then put the converted text back into your document in a Unicode font. Read every line afterwards, because a converter can change a few letters.
• Or retype the text in Mangal or Nirmala UI on a computer, or with a Hindi keyboard on your phone.

Such converters are usually websites, and you can't always tell whether one sends your text to a server. If the text is private, retype it instead. Once the document is Unicode, convert it again.

Converting a Hindi or Tamil document on your phone

1. Save the document to your phone. A file that is only inside a WhatsApp chat or an email isn't saved yet. Download it, or use Share in WhatsApp to save it to your phone's files or Drive.
2. Open Word to PDF, tap Choose Word files and pick the document.
3. Keep the screen on while the row works. Browsers slow down or pause pages you aren't looking at.
4. Read any note under the result. It tells you about an old font, a script that couldn't be drawn, or a chart shown as a grey box.
5. Download and rename. The PDF takes your document's name. If that name is in Hindi letters, give the PDF an English name before you upload it, as some websites may refuse other letters.

How much data it uses

The page downloads only what a document needs. In our tests, a first document in English downloaded about 0.16 MB (a leave letter in Times New Roman) or about 0.2 MB (a resume in Calibri), converter included. Indian scripts need more, because of the text engine and the fonts. The first document downloaded about 0.41 MB for a Tamil notice, and about 0.54 MB for a Hindi office order with English words, because Hindi needs HarfBuzz (about 0.2 MB) and its own fonts. The fonts for one script, regular and bold together, range from 53 KB for Gurmukhi to 191 KB for Bengali; the guide's table of scripts has each one. On a slow connection your first document also waits for its fonts.

How long it takes

On our test computer, the Hindi office order and the Tamil notice each took 0.1 to 0.2 seconds. With the processor slowed down 4 times, roughly a mid-range phone, they took 0.7 and 0.5 seconds. We didn't test a real phone.

English words, digits, bold and italics inside Indian text

Real documents mix scripts: a Hindi order with English words, a Tamil notice with a date in ordinary digits, a table with names in two languages. Here is what to expect:

• English words inside Indian text are fine. They are drawn like any English text, with the open font that has the letter widths of the font they are in, as the fonts part of the guide explains. English in Nirmala UI is drawn with Calibri's widths.
• Digits can be ordinary ones (0 1 2) or the script's own (like १ २ ३). Our test notices had both.
• Bold Indian text is drawn bold.
• Italic Indian text is drawn upright, because the Noto fonts for these scripts have no italic style.
• The letters look different from Mangal or Latha. The PDF uses Noto Sans, so the shapes are Google's, not the ones Word showed. A document in a decorative Indian font looks plainer.

Where lines can break in other places

We compared our PDFs with ones made by LibreOffice, which is not Word but also lays out Word files. The score out of 100 says how much of the print on each page landed within 1.5 mm of LibreOffice's. Bengali scored 99.9, Gujarati 100, Odia 100, Tamil 99.1 and Malayalam 95.9. Punjabi scored 89.4, Kannada 85.5 and Telugu 77.3. In those three, a few lines broke one word earlier or later than in LibreOffice, mostly lines with English words in brackets, so the lines after them sat a few millimetres higher or lower. The words were all there and in order.

What this means for you: if your document must fit on exactly one page, such as a form or a notice, look at the bottom of the PDF. A line that moved can push the last few words onto a second page. Leave a little room at the bottom of the Word file when you can.

Check the PDF the way a reader of that language would

Open the PDF and zoom in. The PDF Viewer opens it right in your browser. Go through it slowly, and ask someone who reads the language to look too if you can.

• Joined letters are one shape. A conjunct such as क्ष or त्र should look like one letter, not two sitting side by side.
• Vowel signs sit where they belong. In हिंदी the sign ि comes before ह, and the dots and marks above the line sit over the right letters. The same goes for the signs in Tamil, Bengali and the rest.
• Names, dates and amounts are right. The PDF repeats the Word file, so a mistake in the file is a mistake in the PDF.
• English words in brackets and table cells are all there. Look at every cell of every table.
• The page count and the last lines are what you expect. See the part on line breaks above.
• The text can be copied. The PDF holds real text, not pictures of pages. Select a line, copy it and paste it into Notes or a chat. If one app pastes it oddly, try another PDF app before you blame the PDF.
• It looks the same on another phone. The fonts are inside the PDF, so a phone with no Indian fonts of its own should show the same letters. Send it to a friend and ask.

When something goes wrong, and what to do

Empty boxes

Part of the text is in a script the converter can't draw yet, such as Urdu, Sinhala, Chinese, Japanese, Korean, Thai or Tibetan. The row says which kind. The text in the nine scripts above is still drawn.

English letters where Hindi should be

An old font like Kruti Dev. Follow the Unicode part above, then convert again.

A wrong spelling in the PDF

It is in the Word file too. Fix it in the document, then convert again.

Italic text stands straight

Indian text in italics is drawn upright. Use bold for emphasis instead if it matters.

Lines break at different words, or the last line moved to a new page

See the part on line breaks. Leave some room at the bottom of the Word file.

A font that isn't in the guide

Indian text is drawn with the Noto Sans font for its script whichever font the document names, so text in another Unicode Indian font should come out the same way. Our tests used the fonts listed in the guide's table of scripts, so check such a document closely.

"An older Word file (.doc)" or "Locked with a password"

The converter can't read either yet. Open a .doc in LibreOffice or Google Docs and save a copy as .docx. For a locked file, ask for an unlocked copy, or open it with the password in LibreOffice and save a copy without one.

A chart shown as a grey box

Charts and SmartArt aren't drawn yet. Pasting the chart as a picture in the Word file fixes it.

The guide lists every note in notes under a result, and the full set of limits in limits.

Using the PDF: portals, merging and keeping the Word file

Size and portals

A PDF made from typed text is small. In our tests, a one-page Hindi office order of 8.0 KB became a 42 KB PDF, and a one-page Tamil notice of 37 KB became a 30 KB PDF. A document with big photos will be bigger. If a portal gives an exact limit, Resize PDF makes the PDF fit under it. The PDF compressor is for when there is no exact number.

An application with your papers

Say you have written a leave application in Hindi in Word, and the office wants it with copies of your Aadhaar and marksheet in one PDF. Convert the Word file, turn the photos of the papers into a PDF with JPG to PDF, then join them in the right order with Merge PDF. Those papers hold private numbers, so it matters that nothing is uploaded at any step.

Keep the Word file

The PDF is for sending. To change it later, change the Word file and convert again. Don't count on turning a PDF back into Word for Indian text. The PDF to Word page says many Hindi and other Indian-script PDFs don't store vowel signs and joined letters in a way that can be read back, and that in two Hindi PDFs made to test this, many words lost them.

No Word on your phone at all? Converting Word to PDF without Office compares the free ways.