Birding by Ear with help from AI
By Margot Bezrutczyk
I’ll probably never know what it was: the bird that sang such complex, liquid song in the thicket of bay laurel that morning on Bolinas Ridge. I tried and failed to come up with a mnemonic, resorting to simile: “It sounds like an Ewok and a robot fighting. It sounds like a computer drowning.” I climbed up some fallen branches hanging off the trail in an attempt to see the bird, but several crows and a wren joined the chorus and began scolding me. I was brand-new to bird identification, armed only with my field guide and binoculars, birding alone during the pandemic, and eventually I simply had to walk away. I found birding in the woods both exhilarating and frustrating, because it was so difficult for a beginner to even know what to listen for.
Then I discovered BirdNET.
BirdNET is a sound ID app developed by the Cornell Lab of Ornithology in partnership with Chemnitz University of Technology in Germany. I’ve been a bit of an evangelist for BirdNET since I started using it, and I was happy to see that similar artificial-intelligence (AI) technology has recently been integrated into the Merlin app as a “Sound ID” feature.
Both apps are free, and both work by recording spectrograms of the sounds detected by your phone. The spectrogram—a visual representation of audio frequency and signal strength or “loudness”—is processed through a neural network that returns one or more predictions based on both the spectrogram and your location. A sound may represent multiple bird species, which the app can identify simultaneously.
You have the option to share your recordings with Cornell’s Macaulay Library, where they’ll be used to improve the quality of IDs in the future. I usually don’t do this because because many of my recordings end with something like, “…it’s just a f*∆king squirrel, again.”
Shortly after downloading BirdNET, I held up my phone towards the cacophony of an oak woodland one April morning, and finally begin to distinguish signal from noise. The first test: I was able to determine that a certain call I’d been hearing for weeks was the song of a Wrentit, an untrustworthy-looking little grey bird I’d seen in my Sibley book and longed to lay eyes on: It was right here all along!
With an idea of what to look for, it wasn’t long before I was able to spot the Wrentit and learn its song by heart.…