Convert a KrutiDev Word File to Unicode (Without Retyping)
A Word document you inherited is set in KrutiDev and you need it in Unicode so it displays on other machines, indexes on the web, or feeds a form that only accepts Devanagari. Retyping the whole thing is the intuitive response and the wrong one. The characters you need are already in the file. What is wrong is only how they are encoded, and a converter fixes that in a paste. This page walks through the full round-trip from Word to converter to Word, and the two things that go wrong if you do not know what to watch for.
Why retyping is not the fix
Word stores the characters. What KrutiDev does is draw Devanagari shapes over Latin code points. A file with the sentence Hkkjr ljdkj holds the letters H, k, k, j, r as ordinary ASCII, with the KrutiDev font instructing Windows to render them as भारत सरकार. Change the font to anything else and the Latin returns, because that is what is actually in the file.
Conversion reads those Latin characters and rewrites them as Unicode Devanagari, real code points rather than glyph substitutions. The Word document then holds ordinary Hindi text that any font can display, and nothing about that requires you to type a word.
Two consequences follow. The KrutiDev font does not need to be installed on your machine for conversion to work, and it does not matter how Word is rendering the text at the moment you copy it. Whether the passage looks like Hindi or like garbled Latin on screen, the same bytes go on the clipboard and the converter gets the same input.
The round-trip in Word
- Open the document and select the KrutiDev text. Ctrl+A works when the whole document is one font. If the file mixes KrutiDev with other content, select only the KrutiDev sections.
- Ctrl+C to copy.
- Paste into the input box of the KrutiDev to Unicode tool and press Convert.
- Copy the Unicode output.
- Return to Word, place the cursor where the source text was, and paste it back. Use Paste Special → Keep Text Only, or right-click and choose the plain-text option in the Paste Options popup that appears next to the pasted block. Plain paste carries whatever style the clipboard picked up and will fight with the Devanagari font you assign in the next step.
- Select the pasted text and switch the font to Mangal, Nirmala UI, or any Unicode Devanagari face you have installed.
That is the entire process. What follows is the set of things that get in the way of it working cleanly.
Two things that go wrong
The first is formatting. Copy from Word carries a lot of clipboard metadata: headings, bold, colour, table cells. A converter's plain text field does not preserve any of that, so what you paste back is unstyled Devanagari. The characters are correct; the layout is not.
For short documents that is a minute of reapplying headings and bold. For long ones, work in chunks. Convert one section at a time, paste it back where it belongs, and set the font on that section only. That keeps the document structure intact and localises any styling fix to the passage you are working on.
The second is mixed encoding. A file that already contained some Unicode Hindi, usually because someone pasted from another source at some point, will look correct in Word and hand the converter two different things at once. The KrutiDev segments convert cleanly. The already-Unicode segments come out as garbled Latin, because the converter reads them as legacy input and applies the wrong mapping.
The fix is to convert only the KrutiDev runs. In Word, select a suspect passage and check the font box. If it reports KrutiDev, or Kruti Dev 010, 020, and so on, it needs conversion. If it reports Mangal, Nirmala UI, or Devanagari MT, it is already Unicode and belongs left alone.
The whole-document Ctrl+A approach works for files with a single legacy font throughout. Anything mixed needs section-level selections.
Tables, headers, and footnotes
These follow the same round-trip but need two adjustments. Tables paste back cell by cell, and the safest approach is to convert one cell at a time so the layout stays put. Selecting the whole table, converting it as one block, and pasting back frequently collapses the table structure into paragraphs.
Headers, footers, and footnotes live in Word regions that are separate from the main body, and Ctrl+A does not include them. Enter each region separately, double-clicking a header to activate it or opening the footnote pane, and repeat the round-trip inside it. Missing this step is the common reason a document appears converted and still shows KrutiDev at the top of every printed page.
Edge cases
Conjuncts and half characters that come out broken
Devanagari joins letters into shapes like क्ष, त्र, ज्ञ, and श्र, and some legacy encodings split those across multiple ASCII characters. Good conversion output has them joined; a converter that drops the joiner produces क् ष or the same letters with a visible dotted circle. If you see that pattern, the underlying issue is at the encoding step rather than in Word, and it is covered on the page about half characters and conjuncts breaking after conversion.
The document names a KrutiDev variant the converter did not expect
Read a few lines of the output before pasting back. If specific words come out wrong while the surrounding text is clean, the source may be a variant other than the one the converter assumes as its default. Check the font box on those runs, and if the name reports a different number, convert those sections separately or standardise the source to the more common variant first.
The document originated in PageMaker rather than Word
Hindi typing in PageMaker predates Unicode support and uses its own habits around legacy fonts. If the source is a PageMaker file exported to .doc, the copy-paste still works, but PageMaker-specific typesetting quirks such as composed characters and discretionary hyphens do not survive Word, let alone conversion. The page on Hindi typing in PageMaker covers the setup side of that separately.
The file is very long
Section by section is safer than the whole thing in one paste, and not only for formatting reasons. Very long inputs make errors harder to spot in the output, and if a single passage does trip up the conversion, working in chunks means you notice which passage and can adjust it in isolation. Ten to twenty paragraphs per pass is a reasonable working size.
File upload is a separate release
This walkthrough is paste-based on purpose. There is no upload-a-.docx path yet on the site, and no fixed date for one to ship. The copy-paste method delivers the same output, works today across every version of Word, and remains the recommended approach regardless of what arrives later.
Variant identification and other legacy-font problems this page sits alongside are indexed on the Hindi font resources page.
Frequently asked questions
No. Conversion reads the characters stored in the file, so the font does not need to be installed on your machine at all. Even if Word shows the text as Latin letters because KrutiDev is missing, the converter still works on those Latin letters and returns Devanagari.
No. A converter reads plain text, so styling does not survive the round-trip. The characters come back correct and unstyled. For long documents, convert one section at a time and paste each back into place, which localises the reformatting.
Mangal and Nirmala UI ship with Windows and render Devanagari cleanly. Any Unicode Devanagari font works, and the choice is aesthetic rather than functional, unlike a KrutiDev document which needs the specific KrutiDev variant it was written in.
Yes. Select the passage, copy it, run it through the converter, and paste the output back where it was. This is the safer approach when a document mixes KrutiDev with text that is already Unicode, since converting the whole thing at once mangles the already-Unicode sections.
They follow the same round-trip but need to be entered separately. Word does not include headers, footers, or footnote regions in a Ctrl+A selection, so open each region and repeat the paste inside it. Tables are safest converted cell by cell so the layout stays intact.
Not yet. The paste-based method covers the same conversion and is what this page walks through. A file-upload path is a separate future release rather than an available option today.