※ vision
A small device sitting near the window that listens to the yard 24/7, identifies any bird species it hears using BirdNET (Cornell Lab's neural network), logs it to SQLite, and sends a push notification when something rare lands. No cloud. No API costs.
※ why
I wanted a telemetry layer for my yard. I wanted to know, when the cardinals show up in March, exactly what time they show up. I also didn't want to pay a SaaS for it. Cloud APIs cost more than the birds.
stack
- ▶BirdNET-Analyzer (Python) for ID — runs on a Mac mini or Pi
- ▶ntfy.sh for push notifications when something interesting hits
- ▶SQLite for the long-tail log
- ▶a tiny Flask dashboard for the daily / weekly / monthly view
- ▶config.py for lat/lon, sensitivity, target species, etc.
- ▶no cloud, no API costs, no telemetry leaving the yard
※ the funny part
this is the project that started the running joke about the bird-poop incident. ask me in person, not online.