Sunday, October 16, 2011

An App to Identify Birdsong

I mentioned this already, but it really deserves its own post. Researchers at the University of Wisconsin are developing an app to identify recorded snippets of birdsong:

Photo: Jeff Miller
For more than a year, Berres (and his graduate students, of course) have been testing and improving the fruit of that inspiration: WeBIRD, the Wisconsin Electronic Bird Identification Resource Database.

Like music-identification apps Shazam and MusicID, WeBIRD allows anyone with a smartphone and a mysterious bird nearby to record the bird’s call, submit it wirelessly to a server and (after a few seconds) receive a positive ID on the species of bird tweeting away within earshot.

“I am amazed at how good it is,” says Berres, who has also used WeBIRD to identify grasshopper species by their clicking calls and frogs by their croaks. “In fact, not only can WeBIRD tell you which species you’re hearing, it’s good enough to identify individual birds from their song.”

For birders, the former qualifies as a reason to rejoice. For researchers, the latter could change the nature of field studies. For the birds, WeBIRD — which hopes to make available to the public in time for the spring migration in 2012 — could be a lifesaver....

Accurate automated analysis of recorded songs could help researchers track the comings and goings of flocks and individuals. Instead of sending students and scientists out into the wild to collect data — collection that could be hindered by variations in hearing, fatigue, biting insects and the very presence of a human being — a research team could venture out periodically to collect recordings of research plots and analyze the results with WeBIRD.

To place a bird call with its species is a chore far more complicated than the music-matching apps....

“When a bird sings, the song itself may have varying amplitudes and frequencies,” Berres says. “It can also speed up a little bit, slow down a little bit. They may throw in a note here or take out a note there.”

Birds also differ their calls throughout the day. And a bird of a particular species on UW–Madison’s lakeside campus may develop an accent of sorts, distinct from birds of the same species living just a few miles away at the UW Arboretum.

The WeBIRD algorithm dices bird calls into time-ordered chunks of frequency and energy, using data-organization techniques more often applied by geneticists to jumbled bits of DNA geneticists to “align temporally misaligned data, working around a lot of the variation,” Berres said.
An app like this was bound to come along sooner or later, and however well it works, it is likely to change birding. If its use becomes widespread among birders, more birders will be able to identify birds by sound. Will this tool be accepted by the birding community? Will identifications by app count the same as identifications by ear or sight? Will the ABA adjust its listing guidelines in light of the new app? These things will all need to be worked out once the app becomes publicly available.