Classroom acoustic analysis from a single photo.
1.5 million children with hearing loss sit in classrooms designed without them in mind. Professional acoustic analysis costs $5,000+. Echo does it in 30 seconds from a phone photo.
Built solo for BearHacks 2026 · Theme: "Break the Norm"
A parent photographs their child's classroom. Echo:
- Detects surface materials (carpet, glass, tile, whiteboard) via Google Cloud Vision API
- Maps room geometry from 4 tapped floor corners
- Runs 38,400 acoustic ray-tracing simulations distributed across DCP workers (16×12 grid × 200 rays)
- Renders a speech intelligibility heatmap — green = clear speech, red = poor
- Generates a spoken voice report via ElevenLabs
- Simulates hearing loss — parents hear "clear hearing" vs "with hearing loss" using Web Audio API frequency attenuation
| Layer | Technology |
|---|---|
| Frontend | Vite + React + TypeScript + Tailwind CSS |
| Auth | Auth0 |
| Image Analysis | Google Cloud Vision API |
| Distributed Compute | DCP (Distributive Compute Platform) |
| Voice & Audio | ElevenLabs + Web Audio API |
| 3D Visualization | Three.js |
| Storage | localStorage |
cd echo
npm install
cp .env.example .env # Add your API keys
npm run devVITE_GOOGLE_VISION_API_KEY=your_key
VITE_ELEVENLABS_API_KEY=your_key
Auth0 is configured in src/main.tsx.
- Google Cloud Vision API — Core input: material detection drives the entire acoustic model
- DCP — Core compute: 38,400 independent simulations distributed across workers
- ElevenLabs — Voice report + hearing loss audio comparison (the demo climax)
- Auth0 — User profiles with persistent scan history
- Best UI/UX — Three.js 3D visualization, interactive heatmap, audio comparison player
- Best Overall — "We gave a parent with an iPhone the computational power of an acoustic consulting firm"
MIT