AssemblyAI Voice Agent Hackathon Submission

Quiet courage.
Invisible protection.
Instant rescue.

When violence strikes, touching a phone provokes assault, and screaming alerts predators. AegisVoice bridges civilian duress and dispatch rescue through hands-free, covert speech intelligence powered by AssemblyAI Universal-3.5-Pro and Claude 3.5 Sonnet crisis triage.

Acoustic Latency
< 1,200 ms

Near-field 16kHz PCM streaming turn-by-turn with keyword recognition.

AI Crisis Triage
< 3.0 s

Structured threat cards, weapons detection, and distress scoring (1-10).

Two-Tier Mesh
2 Tiers

Tier 1: Private security & vigilantes (2m ETA) • Tier 2: Government police 112.

RAM Footprint
~160 MB

Unified Next.js 14 + FastAPI server with zero-CORS single-URL deployment.

Dual Operational Surfaces

Unified deployment under single domain

Civilian Mobile Escort Radar

Route: /

Mobile progressive web app engineered for hostile street environments. Enables hands-free audio streaming with screen wake locks and covert decoy protection.

  • 1.2s Press-and-Hold: Prevents pocket misfires while arming active escort
  • Covert Duress Phrase: Trigger silent beacons while pretending to comply
  • Stealth Camouflage: 0% brightness blackout or functional calculator decoy
  • Off-Thread AudioWorklet: Zero-jank downsampling to 16kHz PCM16
Mobile-optimized interfaceLaunch Civilian Escort

Dispatcher Command Center (CAD)

Route: /dispatch

Mission-control console for security operators, private patrol networks, and emergency agencies. Provides instant acoustic context, telemetry, and 1-click dispatch.

  • Two-Tier Action Matrix: Dispatch nearest private agent (2m ETA) or escalate to police
  • Real-Time Waveform: HTML5 canvas visualizer listening to victim's pocket audio
  • Streaming Transcript: Sub-second word arrivals with danger keyword badges
  • Tactical GIS HUD: Satellite telemetry coordinates with 1-click Google Maps launcher
Desktop 3-column tactical CADLaunch Dispatcher CAD

Under The Hood: Speech Intelligence Stack

How AssemblyAI powers the hands-free defense workflow from ambient audio to responder routing:

1Universal-3.5-Pro

Connects over WebSocket with near-field voice focus, filtering generator hum and street noise while streaming 16kHz PCM audio turn-by-turn.

2Claude 3.5 Sonnet Gateway

The instant distress is detected, ambient speech transcripts are analyzed in under 3 seconds to extract structured JSON (threat level, weapons, tactical brief).

3Embedded TinyDB & SMS Relay

Local file-backed persistence without cloud database costs, coupled with instant 5-contact emergency SMS broadcast with live Google Maps telemetry.