Technology

The Architecture of Listening

Three sensing layers. One mesh network. Zero central servers.

Sensing Layers

From Phone to Planet

Layer 01

Human Sensors

Open the web app and use your microphone. Your phone captures ambient audio, segments it into 3-second windows at 48kHz, and runs species identification in real time. RMS gating ensures only meaningful audio reaches the classifier. Install as a home screen app on iPhone or Android — no app store required.

Layer 02

Acoustic Nodes (Chirp Stations)

Dedicated listening stations running on laptops or Raspberry Pi. Always-on BirdNET inference with continuous audio capture. Each node generates a persistent Ed25519 keypair and signs every detection at the point of creation. Deploy one in your garden and it joins the network automatically.

Layer 03

AI + Mesh Network

Detections propagate across a Reticulum mesh network via heterogeneous transports — WiFi, LoRa, internet, or any combination. Multi-node consensus validates observations. Individual birds receive unique IDs from their vocal fingerprints. No central server required. The network is the intelligence.

Evidence Protocol

CVP-1.1

The Chirp Validation Protocol scores every detection across 9 evidence dimensions. Each signed with Ed25519 at the point of creation. Multi-node consensus across independent observers strengthens evidence beyond what any single node can provide.

Acoustic Confidence

BirdNET model classification probability

Temporal Plausibility

Species expected at this time and season

Spatial Plausibility

Species expected in this geographic range

Signal Quality

Audio SNR, clipping, and background noise

Multi-Node Consensus

Independent corroboration by multiple observers

Vocal Fingerprint

Individual bird re-identification confidence

Behavioral Context

Vocalization type: song, call, alarm, contact

Cryptographic Attestation

Ed25519 signature and provenance chain

Environmental Correlation

Weather, lunar phase, dawn chorus timing

Infrastructure

The Mesh

CORROGO uses Reticulum — a cryptography-based networking protocol designed for resilient, decentralized communication. Detections propagate across any available transport: WiFi, cellular, LoRa radio, packet radio, or wired internet.

Each detection message is 383 bytes. Small enough to traverse any link, from high-bandwidth fiber to low-bandwidth LoRa. The network routes around failures automatically. No central server. No single point of failure. The mesh is the infrastructure.

NODE NODE NODE
By the Numbers
0 Species Recognized
0 Evidence Dimensions
Ed25519 Cryptographic Signing
0 Mesh Message Size
Open Architecture

Built for Research

Every detection is exportable in standard research formats. No lock-in. Your data belongs to you and to science.

eBird Format

Compatible with the Cornell Lab of Ornithology's global birding database. Standard checklist export.

Darwin Core

The biodiversity data standard used by GBIF and natural history museums worldwide.

MoveBank

Animal movement research format. Track migration patterns and individual movements over time.

REST API

Full API for detections, evidence scores, validation reports, contributor data, and real-time WebSocket streams.

Deploy

Run a Chirp Station

A Chirp station is a dedicated acoustic node. It runs on any Linux machine — a Raspberry Pi in your garden, a laptop on your balcony, a server in a nature reserve.

Raspberry Pi

1
Flash the image

Download the Chirp station image and flash to SD card

2
Connect a microphone

USB microphone or I2S MEMS mic for always-on capture

3
Power on

The node generates its Ed25519 identity and begins listening automatically

Any Linux Machine

1
Install

Run the install script: curl -sL chirp.corrogo.com/install | sh

2
Configure

Edit /etc/chirp-node/config.toml for your location and preferences

3
Start

Enable the systemd service: sudo systemctl enable --now chirp-node

Start Listening

Try the app, deploy a station, or read about the vision behind the network.