Chirp 5: Newsletter of 'A sound Idea': acoustic bird monitoring September 2011

by Sarah Lloyd

There was obviously going to come a time when the listening part of the ‘Sound Idea’ project would become overwhelming. For the past several months I have been working on other projects and I have not had much spare time to devote to the task. (Please keep in mind that this is an unfunded project.)

Please do not see this as a reason to stop recording. All recordings contain valuable information and can be analysed at some future date – either by me or by others.

There are exciting advances in the interpretation of digital recordings. Whereas several years ago it seemed that it would be many years before a computer could interpret the recordings, much of this work is now being done using the vast knowledge of ‘citizen scientists’.

The following is taken from a Queensland University of Technology press release:

A Queensland University of Technology (QUT) research team is helping ecologists dramatically slash the time and cost of following the twitter of native birds in order to monitor the effects of climate change.

Until recently, Australian ecologists wanting to identify the presence or absence of wildlife to take the temperature of the changing environment have had to spend days in isolated regions, watching and listening for species in all kinds of weather.

The QUT team, working with the QUT Institute for Sustainable Resources (ISR), gave them a helping hand by developing automated acoustic sensors, placed in the bush to record environmental sounds which are then transmitted to an online digital library.

But ecologists still faced the prospect of sifting through many hours of recordings and isolating bird calls from other sounds like wind and rain - until now.

The QUT team has taken the inspired approach of developing high-tech software and combining it with 'people power' to quickly run over the data and determine the number and type of birds on the recording.Y 'The software filters through the audio and isolates the parts where it can identify potential species amid the cacophony,' said QUT PhD researcher Jason Wimmer.

'People are much better at identifying species than computers, so we leave the final analysis up to the human brain. We post the audio segments our software has identified as containing potential species online and ask the birdwatching community to have a go at telling us what they are.'

Early trials have had fantastic results. Roughly twice as many species have been detected using this approach than traditional surveys with people in the field.

In one recent experiment, while trained observers were able to detect 35 bird species using traditional bird survey methods, the acoustic sensors and software employed by Wimmer picked up 61 birds in the same area.

'We can record the data using inexpensive MP3 recorders or we can upload it via the 3G network,' Wimmer said.

'The most expensive part was analysing the data. But we can reduce this cost by using a 'citizen science' approach and harnessing the enormous resource of birdwatchers - which I'm told is one of the fastest growing hobbies in the world.'

Wimmer and his team verify the expertise of the 'citizen scientists' analysing the data by asking them to sit a test. They then select five people to analyse the audio and accept the majority view.

'We hope to eventually have these acoustic sensors placed all over Australia continuously streaming live data. By speeding up the analysis of that data, ecologists will get a deeper and more current snapshot of climate change,' said Wimmer.

'Birds are indicator species which are sensitive to changes in the environment. If we can find out if there are changes going on in the composition of species, we can respond quickly to ecological alarm bells.'

This project was discussed on Radio National’s Bush Telegraph on August 28th 2010.

Page URL:

Back to top

110906-1, 9, 22, 205, 4083