Listening to nature’s songs

Two green budgerigars nesting in a tree hollow.

The Aussie Bird Count Week is an annual event run by BirdLife Australia. Anyone can participate by observing and counting nearby birds! In this activity, you can learn how to identify birds around you by listening to them using BirdNET, an app developed to quickly identify birds based on their songs.

Safety: You will need to go outdoors for this activity. Wear closed shoes, be sun smart and ask an adult to come with you. Observe wildlife from a distance and do not disturb them.

BirdNET was developed by the K. Lisa Yang Center for Conservation Bioacoustics, Cornell Lab of Ornithology, Cornell University, USA, and the Chemnitz University of Technology, Germany. You can learn more about the app and their research on the BirdNET website.

You will need

A smartphone connected to the internet
Walking shoes

What to do

Search for “BirdNET” on the App Store or Google Play and download the app. Ask a parent for help if you need.
Once you have downloaded the BirdNET app, go to a place where you can hear birds, such as your backyard, a local park or reserve. Ask an adult to come with you if you are exploring outside your home.
Credit: BirdNET, Cornell University, Chemnitz University of Technology

Once you hear a bird call, open the BirdNET app and tap on the “Record” tab on the bottom left to begin recording. You should start seeing a yellow-blue sound graph as the recording moves along.
Credit: BirdNET, Cornell University, Chemnitz University of Technology

After the bird call has finished, you can press the pause button to stop the recording. Then, listen to your recording to find the audio where you think the bird call was the clearest. Next, use your finger to drag and select this part of the recording on the sound graph. Ideally, you should have at least 3 to 5 seconds of audio for analysis.
Credit: BirdNET, Cornell University, Chemnitz University of Technology

Press the ‘Analyse’ button and find out which bird the call belongs to! The app will give you its confidence of the accuracy of the identification, and you can click on the list of options that appears to figure out which bird made the call.

Troubleshooting

Mobile phone screen reads; Sorry we are not able to confidently identify any bird species. — *Credit: BirdNET, Cornell University, Chemnitz University of Technology*

Did you get an identified bird, but with a “Highly uncertain” or “Wild guess”, or no results at all?

It can be tricky to get a good recording, especially if there are multiple bird species calling together. The BirdNET app can only identify one bird species at a time, and if your audio recorded multiple species, you would probably need another recording for the app to confidently identify a bird.

List of birds identified in your area; magpie, swallow, galah, willie-wagtail and grey shrike thrush. — *Credit: BirdNET, Cornell University, Chemnitz University of Technology*

You can also click on the menu tab on top left corner in the BirdNET app and select “Explore your area” to learn about the common birds that are likely to be in your area. You can individually click on each bird and tap on the “Macaulay Library” or “eBird” tabs (found at the bottom of the screen) to listen to other people’s audio and video recordings of a bird. You can try to listen to a particular birdsong and see if you can match it to a bird around you!

What’s happening?

All birds have their own songs. Bird watchers, people who observe and listen to birds, can learn different birdsongs and match them to different birds. Memorising and identifying birdsongs is no easy feat, especially when there’s so many different birds!

But what if you could see birdsongs and their patterns? Researchers who study birdsongs can use a method called a spectrogram to do this. A spectrogram is a graph of a bird call based on things like the frequency (pitch), duration or loudness of each note in a birdsong, similar to a music sheet. Bird watchers often memorise bird songs by looking at these patterns in different spectrograms. You would have seen a the yellow-blue spectrogram on the top panel as you recorded bird calls in BirdNET.

BirdNET uses artificial intelligence (AI) to learn about different bird calls and their unique spectrograms. Then, by matching the patterns in the spectrograms to data on common birds in an area, the app can help bird researchers quickly identify which birds have likely appeared in an area, rather than listening to hours of recordings. Today, BirdNET’s AI has learnt to identify around 3,000 of the world’s most common birds.

Tell us in the comments which birds you’ve found in your area! You can also join the Aussie backyard bird count challenge once you’ve learnt to identify some common birds in your area.

Blog