Sentinel 911

An AI-powered emergency dispatcher system designed to handle multilingual calls with real-time transcription and automated reporting.

Comment

Inspiration

SF 911 gets emergency calls in different languages. Their dispatchers may not be equipped with all languages and the translators might not be available leading to precious time being wasted. Sentinel 911 solves the problem by transcribing the calls in real time and generating the report for human dispatchers for further action while also communicating to the callers.

What it does

Multilingual Support: The system supports Tagalog, Spanish, English, Mandarin, and Hindi. It uses Gemini to detect the caller's language and respond fluently in that same language to keep them calm and gather information. Real-time Transcription: Integrated with the Modulate Velma-2 Streaming API. It captures audio from the microphone, streams it via WebSockets, and receives real-time English transcriptions. AI Interaction: A Gemini-powered agent acts as the dispatcher, professionally handling the conversation and ensuring critical details (location, nature of emergency) are extracted. Automated Incident Reporting: Upon ending a call, the system automatically generates a structured Incident Report in English, summarizing the emergency, urgency level, and caller details. Dispatcher Forwarding: Reports can be forwarded to a human dispatcher (simulated via a backend API call) at the specified email. Mission Control UI: A professional, high-density dashboard built with Tailwind CSS and Framer Motion, featuring live transcription feeds, audio activity indicators, and incident management panels.

How we built it

Technical Details: Frontend: React 19, Tailwind CSS, Lucide Icons, and Framer Motion. Backend: Express.js server for handling API requests and serving the application. AI Models: gemini-3-flash-preview for conversation logic and report generation. Modulate Velma-2 for high-accuracy streaming transcription. Web Speech API for simulated text-to-speech feedback to the caller.

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Sentinel 911

Built With

  • express.js
  • framer
  • gemini-3-flash-preview
  • lucide-icons
  • modulate
  • motion.
  • react-19
  • speech
  • tailwind-css
  • text-to-speech
  • velma-2
  • web
Share this project:

Updates