BirdNET

BirdNET is a project that provides automated detection and identification of bird species from bird sounds. A Shazam for birds. And, when linked with Flickr, I get to see an image of how the species looks whenever a bird is identified.

I’ve had a BirdNET station running for a few months now. It was quick to setup my own via BirdNET-pi thanks to the detailed guide. Just needed a dedicated Raspberry Pi, USB microphone, weather protection, and an outdoor outlet. My station is part of the public “living library of bird vocalizations” at https://app.birdweather.com/. The site puts your station on the map and shares it’s audio feed publicly. Its interesting to see and hear what birds are in various places. Note there are aggressive privacy options to auto purge audio if any human sound is detected, and processing is all done on device. Its amazing how sometimes barely audible recordings can lead to detections.

At scale I’m sure this network can provide a wealth of information for bird researchers to analyze bird migration patterns and population changes. However, at the moment there are very few local stations showing up on BirdWeather. To my surprise there are none in SF, none in East Bay, and none inland. It makes a bit more sense after you learn that the total number of stations globally is only in the hundreds. We need more BirdNET stations!

There are also great phone apps for on-demand identification — BirdNET, Merlin, and BirdWeather. The last of which also shares on the BirdWeather map so that community can discover point in time detections. Highly recommend installing and using it as there are no costs or overheads like running your own station. Low commitment, free, and fun.