अपनी भाषा में। In your own language.
Choose from 46 languages, by name or by where you live. All 22 Eighth Schedule languages are here, along with regional and tribal languages from every state and union territory. Every language is written in its own script, with self-hosted fonts.
Every language, in its own script.
Reviewed checked by a native speaker. Draft key parts translated, being reviewed. Coming soon shows the nearest language until a speaker helps us.
Find yours by where you live.
All 28 states and 8 union territories, from Ladakh to Lakshadweep. Within each place, the most widely read language comes first.
North
Jammu & Kashmir
Ladakh
Himachal Pradesh
Punjab
Chandigarh
Haryana
Delhi
Uttarakhand
Uttar Pradesh
Central
Madhya Pradesh
Chhattisgarh
East
Bihar
Jharkhand
Odisha
West Bengal
North-East
Assam
Arunachal Pradesh
Manipur
Meghalaya
Mizoram
Nagaland
Sikkim
Tripura
West
Rajasthan
Gujarat
Dadra & Nagar Haveli and Daman & Diu
Maharashtra
Goa
South
Karnataka
Kerala
Tamil Nadu
Puducherry
Andhra Pradesh
Telangana
Islands
Andaman & Nicobar Islands
Lakshadweep
Type “dhanyavaad”, get “धन्यवाद”
Phonetic typing for 11 scripts converts each word as you finish it. It runs entirely on your phone: no keyboard app, no cloud service, nothing you type leaves the device.
In the app: 19 languages, scam warnings in 11
The on-phone scam rules understand English, Hindi, Hinglish, Bengali, Tamil, Telugu, Marathi, Gujarati, Kannada, Malayalam and Punjabi. Odia, Assamese and Urdu warnings currently fall back to English while the rules are written.

Why most say “Draft” or “Coming soon”
A bad translation in a safety app is not a small problem: a warning that reads oddly gets ignored. So we label every translation that a native speaker has not yet reviewed as a draft, in its own script, inside the app. An automated test stops us from quietly marking a language as reviewed. On this website, languages we have not drafted yet (such as Santali, Bodo, Khasi or Tulu) show their nearest widely read language and say so plainly, rather than showing a poor machine translation.