The bandwidth problem
Continuous audio is enormous. A network of hydrophones or forest sensors streaming raw sound to the cloud would be impossibly expensive and often impossible full stop — there's no connectivity in the deep ocean or remote forest.
The edge answer
WAVE runs inference on the device. Sub-2MB TinyML models detect species presence locally and send only compact results — not raw audio. That flips the economics of monitoring:
- Offline operation at remote stations.
- Real-time alerts without a round-trip to a server.
- Years of coverage on modest power and storage budgets.
Why it's a big deal
Conservation runs on coverage. The more places you can listen continuously and cheaply, the faster you catch a poaching boat, a species decline, or a shifting migration. Edge AI is what makes planet-scale listening realistic rather than aspirational.
