For years, naturalists and birders have been clamoring for an app like BirdGenie™. There are already apps that identify music, so why not create one for bird songs, too? It sounds easy, but the reality is a lot more complicated.
Programs like Shazam and Soundhoud are excellent at identifying a specific piece of music. But the problem those programs solve is relatively simple. Music is completely consistent – the song in question is always an exact match to the song in the database. Bird song, unfortunately, is not so easy. There are regional and individual variations on every song. Paired with changing environmental conditions, every song that BirdGenie™ analyzes is different. So how did BirdGenie™ solve this complex and daunting problem?
The genesis of BirdGenie™ started several years ago, even before the inception of The Warbler Guide. For years, bird sounds have been described with subjective phrases like teakettle-teakettle or sweet sweet sweet I’m so sweet. Computers are objective, not subjective, so in order to lay the groundwork for an app that could identify songs, Tom Stephenson and Scott Whittle first had to come up with a universal and objective system for describing songs they could then “teach” a computer program.
Some of this work came to fruition in The Warbler Guide, where Tom and Scott presented a complete set of objective definitions that opened up a whole new world of analysis and identification…something unique and revolutionary in the history of field guides.
To create BirdGenie, it took a lot of searching to find the right person to help execute these ideas. After pursuing a few paths, Tom found Stephen Pope. Stephen is a uniquely talented app developer and sound specialist, and an expert in distinguishing sounds from background noise. His experience at Xerox PARC, Stanford’s CCRMA, and UC Berkley’s CNMAT programs, came into play as he tackled to the complex problem of teaching a computer to identify and then analyze bird song.
With Stephen’s specialized skills, and using The Warbler Guide as a foundation, the team perfected a system that analyzes and defines sounds. BirdGenie™ uses hundreds of analytic parameters for each song, some of them undetectable to the human ear, to find the common characteristics in every song. Rather than having to learn each individual iteration of a song for one species, BirdGenie™ can find the critical threads that runs through them all, allowing it to identify song variations that it hasn’t previously encountered. That unique system is patented and promises to bring great accuracy to song identification.
Another very exciting aspect of BirdGenie™ is with its partnership with the Cornell Lab of Ornithology. By sending anonymous recordings from BirdGenie™ users across the country, scientists hope to learn more about bird song variation and function, an area is which is understudied. This is a whole new set of data that may wind up being a very valuable tool in the study and conservation of bird species.
In short, we’re very excited to be producing BirdGenie™! The app is so simple to use, it is easy to forget just how much is going on beneath the surface and how much development was necessary to get the app to work. The unique skills of our team have come together in a special synergy that’s thrilling to experience and the resulting app is functional, accurate, and, frankly, fun to use. We look forward to sharing BirdGenie™ with you and welcome your feedback.