Dhee Indic Transcriber

Description

A Malayali who never learnt Devanagari can read a Sanskrit stotram in Malayalam letters. A Tamil reader can follow a Hindi post in Tamil script. A Bengali can read a Telugu page in Bengali.

Dhee Indic Transcriber puts a slim bar at the top of every page with two controls:

• Read in — a dropdown of scripts. The reader picks one, and every piece of Indic text on the page is rewritten in it, whatever script it was written in. Text in English is never touched.
• Sanskrit mode — checked by default. Every letter keeps its own vowel, as Sanskrit is read: राम becomes Rama, कमल becomes കമല. Uncheck it to read Hindi, Marathi, Bengali, Gujarati, Odia and Punjabi the way they are spoken today: Ram, കമൽ.

The bar stays on screen at every size, from phones to wide desktops. The reader’s choice is remembered in their browser and applies to every page they open, so they choose once.

This is transliteration, not translation: the words stay the same, only the letters change.

How it works

The conversion runs in the reader’s browser, using a JavaScript port of DheeYantra’s SreeVidya engine. Each word is broken into its sounds and the sounds are written out in the chosen script, so any supported script converts to any other.

• Your posts are never changed. Turn the plugin off and nothing is left behind.
• Nothing is sent to a server, and every visitor gets the same HTML, so page caching keeps working.
• Text added after the page loads — comments, infinite scroll, anything a script adds — is converted as it appears.
• Code, form fields and anything marked notranslate are left alone, and you can name other parts of the page to leave alone.

Accuracy

The engine was checked against the original Java SreeVidya over 1,609 lines of real text and 3,989 lines of random characters, in every script and both modes. It gives the same output everywhere except where it deliberately does better: it keeps the chandrabindu (ँ) in scripts that have one, writes Gurmukhi text without stray Devanagari letters, reads Bengali, Odia, Assamese and Gurmukhi vowels and nukta letters correctly, writes Malayalam as Malayalam is written (അമ്മ, ദില്ലി, കല്യാണം, ശങ്കര, പഞ്ച — not അംമ, ദിൽലി, കൽയാണം, ശംകര, പംച), and never loses a sentence to an error.

Some limits come from the scripts themselves. Tamil has one letter for k, kh, g and gh, so text converted into Tamil cannot be converted back exactly; the plugin always converts from the original, so a reader moving between scripts loses nothing. Roman output is phonetic (Namasthe, Rama), not English spelling.

Credits

• The plugin, and the transliteration engine in it (a JavaScript port of SreeVidya), are Copyright (C) 2016-2026 DheeYantra Research Labs Private Limited, licensed under the GNU General Public License version 2 only (LICENSE.txt).
• Gayathri font: Copyright 2023 The Gayathri Project Authors (https://gitlab.com/smc/fonts/gayathri), SIL Open Font License 1.1.