Can you pick a cockatoo’s call from the scree of galah? Artificial intelligence can — and researchers are using that to their advantage.
Queensland University of Technology (QUT) researchers and Google Australia have launched a first-of-its-kind search engine that analyses and identifies wildlife sounds.
It’s hoped the new tool will allow them to track where some animals end up after bushfires or other natural disasters, and to help inform conservation decisions about species and ecosystems.
Before we get into it though, let’s see if your skills can match the machine’s.
The bird calls you just heard form part of the new audio search engine’s database, which includes millions of hours of wildlife audio collected by the Australian Acoustics Observatory (A2O).
A2O has a network of recording devices operating at approximately 360 sites across the country, but human researchers simply don’t have the time to manually identify and analyse all of the sounds that they collect.
Paul Roe, a professor at QUT’s School of Computer Science, said A20 Search is an answer to that problem.
“In the acoustic observatory, we’ve got so far 300 years of data, and that will increase to over 1,000 years of data,” he said.
“If you’ve got one person, it’s going to take many lifetimes for them to listen to all of that data if they were to do that manually.”
Using Google AI and automatic audio recognition, users can upload audio recordings of an animal species, find similar sounds across the database, filter by location and date, and download the results.
“So the neat thing … is that with just a single call, we can search in real-time across all this data to find out essentially where in Australia these species [are] and when they [were] there,” Professor Roe said.
“This has never been done before — we’ve created essentially a Google search for the acoustic observatory.”
Using sounds to understand ‘how things are changing’
Researchers hope the audio search engine will yield important insights into where some animals go after bushfires and other natural disasters, and to better understand the impact of climate change and the spread of invasive species.Professor Roe said it would also be used to track migratory species and those that are endangered, which are typically harder to locate.
“There’s a bit of a catch-22 situation that often the species that we’re most interested in, we may not have a lot of example [animal] calls for that species; it may be some [are] critically endangered species,” he said.
“So we can use [the data] to answer a lot of questions around how things are changing.”
Professor Roe said the new tool could have real-world impacts for conservation efforts.
