On a recent rainy morning in Brooklyn, Tom Stephenson, smiling and serious, dressed in khaki quick dry and a baseball cap with a Cerulean warbler insignia, stepped into Prospect Park, 526 acres of sweeping lawns, forest, lakes, and winding paths. Through layers of car honks, distant jackhammers, and chatty joggers, Stephenson heard a delicate, high pitched trill. “Cedar waxwing,” he said. Walking beneath a canopy of dogwood trees, he called out, “Robin, robin, catbird.” The catbird was mimicking other birds, he said, stringing together a jumble of song fragments. It sounded like a distracted robin. “A hairy woodpecker!” he exclaimed. “Another robin, a yellow warbler.”
For Stephenson, bird watching is bird hearing. Over the past 40 years, he’s tuned his ear to the canopy. Stephenson is famous in the warbler world as a guide and author. In 2012, he won a category in the World Series of Birding, a team competition to identify as many birds as possible in New Jersey over 24 hours. A few years later, he set the United States record for Photo Big Day, another team competition, photographing 208 species in one day (many which he tracked by ear). Now, he is more interested in sharing his ears with others.
Stephenson’s latest project is BirdGenie, a “Shazam for birds.” Shazam is an app for identifying a song by recording a short sample of it. It creates a digital fingerprint of the song, matches the fingerprint to a database of songs, and reports the artist and song. BirdGenie, due to be released this fall, will do the same for birds. So far, it can identify a few hundred of the most common North American species.
In Prospect Park, Stephenson held his phone to a tree to record a bird call. The app displayed a sonogram of sound—a plot of pitch changes. It filtered low frequency noise from passing airplanes and voices. There were two slides downward in pitch followed by a dozen evenly spaced steep rises. On the next screen, BirdGenie identified potential matches, including the species Stephenson had already identified, a Northern cardinal. A bit later, Stephenson recorded a house wren. The sonogram was a staircase of sound, a series of chirps gradually declining in pitch.
Stephenson began training his ear after a birdwatching trip to Ecuador in the ’80s. “There are lots of birds you want to see that are very hard to find—wrens, ant thrushes—and they call quite a bit,” he told me. “It really pissed me off that I didn’t know the songs!” One of the challenges in training your ear is that opportunities to practice are rare. “If you’re just learning in the field, you only hear them a couple times, and then you don’t hear it at all for a year,” Stephenson said. “Birds don’t sing much in the fall.”
Identifying a bird song is a more difficult problem than a human song.
At that time, there was no good resource for a birder seeking sound tracks. You could get cassette tapes from a guy named John Moore, Stephenson said. But the tapes were only available for a few regions. So Stephenson began song-collecting, gradually compiling an encyclopedia of recorded calls. He sketched the shapes of the songs, the rises and falls of pitch, as a study aid. Now, Stephenson draws upon his own library as well as crowdsourced databases housed by the Cornell Lab of Ornithology and xeno-canto, an online collection.
Stephenson estimates he’s learned over 3,000 calls. For any birdwatching trip, he often learns several hundred new songs. But first he had to learn how to study. “You can’t just learn 200 songs at once,” he said. Memory books written for high school teachers, college students, and med students, he said, “all say the same thing: Learn about six or seven calls at once.”
Stephenson developed a visualization technique for memorizing bird songs. In an addendum to 2013’s The Warbler Guide, which he wrote with Scott Whittle, he explained that abstract images were key. To memorize the buzzy song of a prairie warbler: “Picture a gently rising prairie field filled with bees buzzing. If you picture yourself walking in this yellow field, like the color of the bird, you will have an even stronger image.” He told me, “The crazier the connection, the better. Your brain remembers crazy things much better than it remembers simple things, actually.”
Most field guides use transliteration to write bird calls as strings of syllables. The golden-cheeked warbler, known as the Texas gold finch, can be heard singing “Zo zu zu zo zu zhray ZHEE,” according to one guide. But transliteration is both unsatisfying and wildly inconsistent. Stephenson surveyed 20 of the field guides on his bookshelves. Each described the warbler’s song differently. “Buzzing d’d’d’d drrr-drruz-zee,” one said. Another: “Bzz Bzz Bzz Bzz bzza-wee”; another: “Tweeah tweeah tweesy.”
Stephenson thought this could be simplified. “I spent a lot of time with thousands of sonograms looking at structure, the phrasing, number of elements, and pitch changes,” he said. Stephenson was a musician long before he was a bird listener. In the ’70s, he played “cello, classical percussion, marimba, some synthesizer, all kinds of weird instruments” on the college music circuit. He stayed in the music business for a while, recording for the Grateful Dead, editing for Phil Collins and, later, managing recording at the Roland Corporation, a major producer of electronic instruments and recorders.
Stephenson categorized his bird song database like a sound engineer, describing each sonogram like a short musical composition. “I had big spreadsheets of this stuff that I was looking at, until I could find rules; things that were consistent.” He built a dichotomous key for song. “The first question is the quality. Is it a buzzy song, clear, or trilled?” Next, does the song increase in pitch, decrease, or stay constant? And finally, how is it phrased? For the golden-cheeked warbler, it’s not necessary to decide if you’ve heard a “Zo zu” or not. Stephenson came up with a simple diagnostic rule to describe the golden-cheeked warbler’s unique sonic footprint. The song was composed of “four buzzy elements and a clear element.” No other warbler song shares this pattern.
“Identifying a bird song is a much more difficult problem than identifying a human song,” Stephenson explained. “Human songs have so many constants—they’re often in the same key, with similar pitch changes, timing between notes, and rhythms.” They don’t vary when played on different sound systems. “Birds are infinitely more variable both in terms of structure, pitch, every parameter you can think of.”
Stephenson said identifying bird songs was more than a technical challenge or game. He remembers visiting friends on Long Island. When he asked what species visited their birdfeeder, they responded vaguely, “Chickadees and some other things.” That afternoon, Stephenson identified 25 species. His friends were thrilled. “Learning birds is not unlike learning about art or architecture,” Stephenson said. “It focuses the eye, helps you see things you never would’ve seen. Knowing the vocabulary of feather parts of a bird helps you see those parts. Without that, you wouldn’t see the detail. It’s the same thing with sounds. Once you have the language of vocalizations, you can begin to hear pitch trends, quality, other characteristics.”
BirdGenie could also help build a resource for biologists. There are 50 million bird feeders in the United States alone. If people begin to use the app, it will “gather a database of vocalizations across the country.”
We walked through an area of Prospect Park known as Midwood, Brooklyn’s oldest forest, and heard the catch of the day, a male Acadian flycatcher. We followed it down damp stone steps and across and across a rock bridge, near an old sign that said “Rick’s Place.” As he walked, Stephenson read a tweet to Siri in hushed voice: “Acadian flycatcher calling at the stairs down from Rick’s Place toward waterfall.” (It’s considered good birding etiquette to update the community about rare finds.) A birder on a bike was already there, scanning the canopy. Stephenson silently nodded at him, an old bird friend.
The first caller was not alone. “Definitely pair of Acadian flycatchers,” @TomProspectBird tweeted, and we followed the calls further into the Midwood. This was excellent news, suggesting that the flycatcher pair could be breeding in Brooklyn. “Now you’re engaging in the incredible birding activity of great anticipation of a cool species! In the rain, too!” Stephenson said. Finally, Stephenson pointed them out. They were discreet olive birds, the size of oak leaves, with black and white wing bars and light yellow bellies. They are birds often heard and not seen.
They’re not cute, though. That word is banned from Stephenson’s walks.
Katharine Walter is an editorial fellow at Nautilus.