Australian bird life is awesome.
We’ve got everything from the astonishing Cassowary through to the Fairy Penguin, the immense NBN-Cable-Chewing-Jerks that are the Sulphur-Crested Cockatoos and yes, even the much-maligned (but secretly excellent) Bin Chicken.
However, all is not well in Australian Bird landscape, with an increasing number of Australian bird species identified as threatened species. That’s not threatened as in “the local cat looked at me funny” (though cats, both feral and domestic are a big issue here) but instead threatened with extinction.
Extinction, if I have to remind you, is bad. You should not welcome the prospect of wildlife extinction in any way.
One way that we can positively move towards protecting more of our feathered mostly-flying chums is to identify the areas where they live, both for signs of their existence as well as signs of predator species, like the aforementioned cats, for example.
That’s where Google, working with the Queensland University of Technology is using AI to help efforts to identify and monitor native bird species. One way you can do this is by identifying their bird calls, but this has historically involved researchers trawling manually through recorded footage.
To give this some perspective, QUT’s A2O project has, since 2019 captured more than 17 million hours of raw environmental audio. That’s one heck of a Spotify playlist, but also a big challenge for accurate measurement that could have a profound impact on the way that wildlife spaces are managed now and in the future.
Google’s role here is part of its Digital Future Initiative, an investment that has already seen it fund research on everything from hearing health to quantum computing.
For Australia’s bird life, Google worked with QUT’s A2O project to use AI and known recordings of one threatened species, the Australian Glossy Black Cockatoo. They built out a smart model that uses machine learning, applying it to that raw audio to identify the distinctive vocalisations of the Glossy Black Cockatoo far more rapidly than with manual searching.
The model is very cool, able to separate out parts of sound that are identifiably just wind, other birds, mammals and insect noises, even when – as is the case for the Glossy Black Cockatoo – their actual calls might be highly intermittent and quite soft to actually hear.
The objective here is to ultimately be able to produce species range maps to allow conservationists to track populations and undergo research tasks more efficiently and effectively as we strive to protect Australia’s immense natural wealth of differentiated species.
What’s more, the model itself is open source, so it’s not a case of Google locking it behind a paywall – though you’d essentially have to be a wildlife researcher for it to make much sense to you to use.
Despite the comments about cats above, I swear, I’m a cat person, by the way. I have three of them. And they live indoors exclusively, because while I love cats, I also love Australian wild life, and I know what cats are underneath the cute. Want to do your part to help save Australian birdlife? Keep your cats indoors — and save a fortune in vet bills while doing so.