The moment we're building for — हर घर की एक सच्ची कहानी
SunoSathi turns your own mobile number into a captioned phone. When someone calls, you read live subtitles of their words — in हिन्दी, ગુજરાતી, English or Hinglish, detected automatically (more Indian languages coming soon) — and reply with your voice, by typing, or by tapping pictures that speak for you. The caller just… calls. No app, no link, nothing special.
Assistive tech usually asks the world to change. SunoSathi doesn't — the caller dials a normal number and talks normally.
Dial one forwarding code (**21*…#) once, and calls to the number your family already knows ring in SunoSathi — with vibration and lock-screen alerts, even when the phone is in your pocket.
The caller's words appear as live chat bubbles in the language they speak — Hindi, Gujarati, English, or mixed. If a language can't be identified, it's written the way it sounds — "kem cho".
Speak, and the caller hears you directly. Or type — they hear your words in a natural voice you choose. Or tap picture tiles that say whole sentences for you. Every way is a real reply on a real call.
Auto-detects Hindi, Gujarati, English and Hinglish mid-sentence — powered by Sarvam AI, built for Indian speech. Marathi, Bengali, Tamil, Telugu, Kannada, Malayalam, Punjabi and Odia coming soon.
Captions land in under a second at natural pauses. A real conversation, not a transcript afterwards.
Call your contacts or any number. They see a normal call from your number — and hear your real voice.
Every call saved as transcript + audio. The doctor's instructions, the delivery address — always there to re-read.
Continuous vibration and lock-screen alerts — even with the app closed and the screen off. Every call gets noticed.
Can't speak? Type — the caller hears your words a second later in a natural voice you choose. The moment you start typing, the caller is told to hold on. One-tap phrases for the quick stuff.
A picture board speaks whole sentences with one tap — हाँ, रुकिए, "मैं घर पर हूँ". Built for sign-language-first users. No typing, no reading needed.
Every call is automatically re-checked for caption accuracy. We publish our numbers to ourselves before anyone else.
Street, TV, kitchen — background sound is learned and silenced before it can turn into junk captions. Speech over the noise still gets written.
Couldn't answer? A Telegram message tells you who called and when — so a missed call never becomes a missed doctor, delivery or friend.
Hard of hearing, not deaf? Keep the caller's voice on alongside the captions — or switch it off for silence. Captions grow to the size your eyes want.
Android app, iPhone app, or just a browser tab. No microphone? No problem — the call simply runs in type-to-speak mode.
"Such a need of the hour! How can one get this tested? I run an NGO for around 70 deaf children and adults… This could be incredibly useful for the deaf community."
"Execution is impressive, but choosing the right problem is what makes this stand out."
"Didn't think of a problem like that could be solved using speech models. Great stuff."
"Possibly useful for people without hearing difficulties also — when you're in a quiet room and don't want anyone else to listen."
We're onboarding early users, families, and NGOs now. It's free during the pilot — we want feedback, not money. Tell us who you are and we'll reach out personally.