Audio Intelligence Engine — Acoustic Distress Detection & Dispatch

HACQUIRE 2026 Tradable Software Asset
Proposed Asking Price: ₹4.00 Cr
Category: Acoustic Intelligence & Voice AI
Integration Complexity: Low (Drop-in REST / Webhook API)

---

1. What This Module Does

Audio Intelligence Engine is a standalone acoustic monitoring and voice distress processing module. It captures real-time audio from edge microphones, performs continuous speech-to-text keyword extraction, detects critical distress terms (HELP, FIRE, POLICE, GUNSHOT, AMBULANCE), matches the distress type to the appropriate response agency (Fire, Police, Hospital), and automatically dispatches alerts.

---

2. Problem It Solves

Victims in distress often cannot type on a phone or reach an emergency button. In noisy campus environments, human dispatchers cannot monitor hundreds of microphone feeds manually. Audio Intelligence Engine solves this by running continuous acoustic listening, visualizing decibel frequency spectrums in real time, and instantaneously triggering automated agency dispatches when emergency keywords are shouted.

---

3. Key Capabilities

• Continuous Live Microphone Sentry: Web Audio API frequency analysis and Web Speech API continuous recognition.
• Dynamic Decibel Spectrum Visualizer: 24-band animated frequency spectrum with decibel intensity tracking.
• Intelligent Agency Routing:
  • FIRE / SMOKE ➔ Fire Station Unit 1
  • POLICE / GUNSHOT / INTRUDER ➔ Police Department Rapid Response
  • AMBULANCE / HOSPITAL / MEDICAL ➔ Campus Hospital Paramedics
  • HELP ➔ Central Campus Security
• Acoustic Simulation Sandbox: Instant manual event trigger for testing keyword classification and webhook dispatch.
• Dual Webhook Dispatch Engine: Concurrently dispatches structured alert payloads to upstream Notification Gateways and Command Centers.

---

4. Architecture

┌─────────────────────────────────────────────────────────────┐
│                  Audio Intelligence Engine                  │
│                                                             │
│   ┌─────────────────┐       ┌───────────────────────────┐   │
│   │ Live Microphone │ ───►  │ Web Audio Spectrum + FFT  │   │
│   │  Audio Stream   │       │ Speech Keyword Recognizer │   │
│   └─────────────────┘       └─────────────┬─────────────┘   │
│                                           │                 │
│                                           ▼                 │
│   ┌─────────────────┐       ┌───────────────────────────┐   │
│   │ Upstream Webhook│ ◄───  │     Agency Dispatch       │   │
│   │ & Notification  │       │     Routing Engine        │   │
│   └─────────────────┘       └───────────────────────────┘   │
└─────────────────────────────────────────────────────────────┘

---

5. Installation

# Clone the standalone repository
git clone https://github.com/2405227-dev/audio-engine-hacquire.git
cd audio-engine-hacquire

# Install dependencies
npm install

# Initialize the local database
npx prisma generate
npx prisma db push

---

6. Environment Variables

Copy .env.example to .env.local:

PORT=3001
DATABASE_URL="file:./dev.db"
NOTIFICATION_SERVICE_URL="http://localhost:3003/api/notify"
SCER_WEBHOOK_URL="http://localhost:3000/api/webhooks/audio"

---

7. Running Locally

# Development server (Port 3001)
npm run dev

# Production build
npm run build
npm run start

Access the Acoustic HUD at http://localhost:3001.

---

8. API Reference

POST /api/voice-detect

• Purpose: Receives live speech detection events, logs them, and dispatches automated agency alerts.
• Request Body:

{
  "detectedKeyword": "FIRE",
  "confidenceScore": 0.96,
  "audioLevelDb": -14.2,
  "sensorLocation": "Zone 4 - Academic Quad",
  "rawTranscript": "Fire in the building help!"
}

• Response:

{
  "success": true,
  "eventId": "evt_clx...",
  "keyword": "FIRE",
  "targetAgency": "Fire Station (Dispatch Unit 1)",
  "severity": "CRITICAL",
  "dispatched": true
}

POST /api/simulate

• Purpose: Simulates an acoustic distress trigger for testing and validation.
• Request Body:

{
  "keyword": "HELP",
  "confidence": 0.95,
  "location": "Block C - Floor 2"
}

---

9. Integration Guide

To send distress detections from any external sensor or Raspberry Pi:

import requests

payload = {
    "detectedKeyword": "HELP",
    "confidenceScore": 0.98,
    "sensorLocation": "Library East Wing",
    "rawTranscript": "Help me please!"
}
response = requests.post("http://localhost:3001/api/voice-detect", json=payload)
print(response.json())

---

10. Example Workflow

[ Voice Distress Shouted: "Fire!" ]
              │
              ▼
[ Web Audio FFT + Keyword Matcher ] ──► Extracts "FIRE" (Confidence 0.96)
              │
              ▼
[ Agency Router ] ──► Routes to "Fire Station (Dispatch Unit 1)"
              │
              ▼
[ Dispatch Payloads ] ──► Sends critical alert to Smart Notification Engine & SCER Radar

---

11. Limitations

• Browser speech recognition uses the Web Speech API (Chrome / Edge / Safari supported).
• Background noise suppression thresholds are configurable via audioLevelDb.

---

12. License / Usage

Prepared as an official HACQUIRE 2026 Tradable Software Asset. Distributed for hackathon asset trading and integration sprints under non-exclusive participant terms.