Suno Sarkar
Just speak government listens.
The problem Suno Sarkar solves
What can people use Suno Sarkar for?
Suno Sarkar is a voice-controlled government services assistant that allows citizens to access essential public services using natural speech in Hindi or English. Instead of navigating complex portals or forms, users can simply ask questions by voice and get instant, easy-to-understand responses
How it makes existing tasks easier
Talk instead of typing
Users can ask questions like:
“Mera gas subsidy status batao”
“Mujhe Ayushman card milega ya nahi?”
No need to read long instructions or fill confusing forms.
Smart AI understanding
The system understands user intent, even with informal or mixed Hindi-English language.
Automatically maps the query to the correct government service.
One place for multiple services
People can access information related to:
Ration card status
LPG gas subsidy
Pension schemes
Ayushman Bharat eligibility
All from a single, simple interface.
Voice + text responses
Responses are given in spoken voice and clear text
Helpful for elderly users, low-literacy users, and visually challenged citizens
How it makes services safer & more accessible
Reduces dependency on middlemen
Prevents misinformation and fraud
Users get direct, reliable guidance
Elderly & rural friendly
Large buttons
Minimal UI
No technical knowledge required
Language inclusion
Supports Hindi and English
Designed for future expansion to regional languages
Real-World Impact
Saves time and effort for citizens
Improves access to government services
Encourages digital inclusion
Makes governance more human, approachable, and transparent
In short
Suno Sarkar turns complex government systems into simple voice conversations — making public services accessible to everyone.
Challenges I ran into
Inconsistent Voice Recognition (Hindi + Hinglish)
Problem:
Voice recognition was inaccurate when users spoke mixed Hindi-English (Hinglish) or had different accents. This caused incorrect text conversion and wrong intent detection.
Solution:
Switched to a hybrid approach:
Browser Speech API for fast transcription
Custom keyword + intent matching layer
Added fallback suggestions when confidence was low
Normalized text by removing filler words and variations
Result:
Significantly improved accuracy for real-world Indian speech patterns.
Intent Detection for Informal Queries
Problem:
Users don’t follow fixed sentences. Queries like:
“Gas ka paisa aaya ya nahi?”
did not match standard intent rules.
Solution:
Built a rule-based + NLP hybrid engine
Grouped synonyms and common phrases
Introduced confidence scoring before mapping intent
Result:
The system became more flexible and user-friendly without needing a heavy ML model.
Performance on Low-End & Mobile Devices
Problem:
Voice processing and UI animations caused lag on low-end Android devices.
Solution:
Optimized UI by:
Reducing re-renders
Using lightweight animations
Deferred heavy processing until voice input completed
Result:
Smooth performance even on budget smartphones.
Lack of Real Government APIs
Problem:
Most government services do not provide open APIs for real-time data access.
Solution:
Created realistic mock APIs based on public documentation
Designed the system to be API-ready for future integration
Result:
A fully working demo with a clear path to real-world deployment.
Summary
The biggest hurdle was understanding how real people actually speak, not how systems expect them to. By adapting our AI to human behavior instead of forcing users to adapt, we made Suno Sarkar truly inclusive.
Technologies used