Software to record, identify animal calls developed by Alberta researchers

A research lab at the University of Alberta is developing new technology allowing researchers to listen in on animals in places researchers might not otherwise have access to.

For five years, the bioacoustic unit at the University of Alberta has been setting up small computers and cellphones hooked up to microphones in far-flung wetlands, mountains and forests of Alberta.

The project got its start when biology professor Erin Bayne was working with industry members in the energy sector to help monitor yellow rails, a small and relatively rare bird, known for singing only in the middle of the night.

"It's a very hard animal to monitor," Bayne said.

Wanting to avoid stomping through the dark wilderness in search of the hard-to-spot birds, Bayne started developing the audio recording technology he still uses today.

Relying largely upon either solar energy or an external battery packs, the audio recording units can be programmed to run at set times and intervals and then hiked in or dropped into place from helicopters.

Once deposited, they are left in place for days, weeks, months or even years, giving researchers the most complete understanding of what animals are up to available to date.

Through years of bioacoustic analysis, Bayne's team eventually determined that, while still rare, yellow rails are more common than previously thought.

'Shazam for animal songs'

Today, the bioacoustic unit is run as a partnership between Bayne's lab and the Alberta Biodiversity Monitoring Institute, a group of individuals and agencies that monitors animal diversity across province.

Its recordings are used to track species movement, migration and population, as well as the impact humans and climate change may be having on animal life.

The lab monitors anything and everything that makes a sound, including migratory birds, owls, geese, bats, frogs, moose, beavers and coyotes.

When the recording period is over, the data is brought back to the lab where experts are occasionally enlisted to identify individual species.

"But the problem with that is we have far outstripped our ability to do it," Bayne said. "If I listened to all the data I've collected in the last couple of years, I (would be retired) by the time I've actually listened to all of it."

Increasingly, the lab is relying on new audio technologies that are able to recognize the call of individual animal noises, much like the smartphone app Shazam is able to recognize songs, Bayne said.

Because so much of the analysis relied on parsing out animal noises from other ambient sound like wind, rain and other animal noises, the analysis software requires more subtlety than the song app.

Bayne's team is currently in the process of training the software to recognize new animals one species at a time.

So far, he said it's working well when species "sing in a perfect world — really close to the microphone and are really isolated with not lots of other things happening."

But he is working on improving that.

Technology always improving

Altogether, Bayne estimates his team listens to only about half a percent of the total audio collected, mainly just to confirm the automated process is working. In the future, the goal is to have theirs computers go through 100 per cent of the data.

The use of bioacoustics has opened the doors for studying many more species than was possible using traditional field techniques, Bayne said, but should ever be expected to replace field research.

"It complements some of the things people are doing and it will help us with a whole bunch of species that've been having to ignore because people can't get to the places the animals are."

Soon, the lab hopes to share their recordings with the public online on its website.