Skip to content

Resonance v4.0.0 — Spatial Acoustic Intelligence Platform

Choose a tag to compare

@theworker02 theworker02 released this 26 Aug 16:57
· 24 commits to main since this release

Resonance v4.0.0 — Spatial Acoustic Intelligence Platform

The first public release of Resonance: an open software platform for distributed acoustic event detection using calibrated sensor arrays, probabilistic spatial inference, and auditable evidence chains.


What Is Resonance?

Resonance replaces traditional "pin on a map" acoustic detection with a measurement pipeline that reports what happened, how certain it is, and why it reached that conclusion.

Pressure Wave → Multi-channel Acquisition → Direction Estimation → Spatial Cell Correlation
→ Environmental Compensation → Classification → Evidence Fusion → Auditable Incident

Every detection includes explicit uncertainty. No single sensor can produce a high-confidence incident.


Highlights

Signal Processing (resonance-sdk on crates.io)

  • VectorWave — direction-of-arrival estimation using GCC-PHAT cross-correlation + delay-and-sum beamforming with calibrated uncertainty intervals
  • WavePrint — perceptual acoustic fingerprinting for cross-sensor event correlation
  • EchoGraph — multi-path decomposition that separates direct arrivals from reflections
  • Acoustic Probability Surface — continuous spatial probability field replacing point estimates
  • ConflictGuard — automatic contradiction detection between sensor observations
  • Scene Health — environmental condition assessment that caps confidence when conditions are poor

Platform (resonance-platform on crates.io)

  • Chronos — precision timing management with GNSS PPS and holdover quality scoring
  • Atmosphere Engine — speed-of-sound correction from temperature, wind compensation for DOA
  • NodeCare — predictive maintenance scoring from health telemetry
  • Spatial cells, cross-node correlation, provenance chains, incident lifecycle

Edge Runtime (resonance-edge on crates.io)

  • Hardware Abstraction Layer (Linux + Simulator backends)
  • DSP pipeline: normalizer → impulse detector → feature extractor
  • Privacy kernel with compile-time prohibited capabilities
  • Ed25519-signed REP observations with replay protection
  • Offline queue with reconnect replay

Cloud Platform (apps/cloud/)

  • Multi-tenant architecture (Organizations → Workspaces → Deployments)
  • 8-role RBAC with 30+ granular permissions
  • JWT authentication with API keys and webhooks
  • Event-driven domain architecture with typed errors
  • Background job framework with exponential backoff
  • Explicit state machines for Node, Incident, Model, and Deployment lifecycles
  • Data retention engine (hot → warm → archive → delete)
  • Feature flag system with percentage rollout and targeting

Design System (packages/surface/)

  • Complete semantic design tokens (colors, spacing, typography, radius, elevation)
  • Dark and light theme CSS variables
  • Motion specification (5 categories respecting reduced-motion)
  • Component interface contracts for 15+ visualization primitives
  • Accessibility spec targeting WCAG 2.1 AA

Hardware Reference

  • VectorNode X1 engineering specification with 30+ formal requirements
  • Manufacturing plan for contract electronics manufacturers
  • Reference BOM, array geometries, and node family definitions
  • Published under CERN Open Hardware Licence (permissive)

Screenshots

Incident Confidence Breakdown

Incident Detail

Every incident shows dimensional evidence scores: classifier confidence, sensor agreement, TOA consistency, signal quality, and environmental model — not just a single percentage.

Simulation

Simulation Output

The simulator creates virtual sensor networks for development without physical hardware. Run resonance simulate --nodes 25 --environment suburban --duration 5m to try it.


Architecture

┌─────────────┐     ┌──────────────┐     ┌───────────────┐     ┌──────────────────┐     ┌────────────────┐
│  VectorNode │────▶│ Spatial Cell  │────▶│   WaveGraph   │────▶│ Probability Field │────▶│ Evidence Fusion │
│  (edge DSP) │     │ (4-8 nodes)  │     │ (propagation) │     │   (uncertainty)   │     │  (confidence)   │
└─────────────┘     └──────────────┘     └───────────────┘     └──────────────────┘     └────────────────┘

Privacy by Architecture

Resonance is architecturally incapable of:

  • Speech recognition
  • Speaker identification
  • Continuous audio recording
  • Keyword monitoring
  • Individual tracking

These aren't policy decisions — they're enforced at the type system and protocol level. Raw audio never leaves the sensor node.


crates.io

All three crates are published:

Quick Start

# Clone
git clone https://github.com/theworker02/resonance
cd resonance

# Build
cargo build --workspace

# Run simulation
python simulator/src/main.py --nodes 25 --environment suburban --duration 5m

# Start the cloud platform
cd apps/cloud && npm install && npm run dev

What's Next (v4.1)

  • First prototype PCB fabrication
  • Hardware-in-the-loop testing
  • Fleet manager application
  • Public benchmark datasets
  • Detector SDK with third-party marketplace

Contributing

See CONTRIBUTING.md for development setup and guidelines.

License

Apache 2.0 (software) · CERN-OHL-P (hardware reference designs)