The goal

Intended Impacts

• Deliver innovative, lower-cost biodiversity monitoring compared to traditional methods
• Build confidence in the integrity of nature repair market outcomes for buyers
• Provide monitoring tools scalable across diverse Australian environments
• Support the broader nature repair and ecosystem services market with reliable, evidence-based biodiversity data

The result

What Has Been Achieved: Automated Bird Call Recognition for Woodland Biodiversity

The research team at the University of New England, in partnership with Gulbali Institute, Charles Sturt University, the University of Queensland, Bush Heritage Australia, and NRM South (Tasmania), has successfully developed Australia's first regional bird call classifier for south-eastern woodlands — one of twelve projects commissioned under the Federal Government's Innovative Biodiversity Monitoring Program.

The Tool

The completed tool uses machine learning — specifically a convolutional neural network built on the Cornell Lab of Ornithology's open-source BirdNET framework — trained over two years by two postdoctoral researchers to reliably identify bird species by sound in the Australian woodland context. It covers 183 woodland bird species across south-east Australia from southern Queensland through to South Australia and Tasmania, including both day and night species.

How It Works

Users upload recordings from passive acoustic recorder sensors, run the recogniser remotely and receive a complete species list with adjustable confidence thresholds. At the recommended confidence levels, the tool delivers highly accurate results with minimal false identifications. A minimum of one week of recording is recommended for reliable outputs, and results can be exported directly to the Atlas of Living Australia.

Proven Performance

Field testing has demonstrated the tool's remarkable capability — at one trial site where expert surveyors identified 60 species over three months, the tool detected 95 species from the same period's recordings, outperforming traditional methods significantly.

Accessibility and Deployment

The tool is now being delivered to the Federal Department of Climate Change, Energy, the Environment and Water (DCCEEW) and will be freely available through their website as a recommended monitoring tool. How-to guides are being developed to support uptake by landholders, Landcare groups, Local Land Services, and regional bodies.

Broader Impact Already Underway

Beyond birds, the approach is being adopted by colleagues building equivalent tools for frogs, bats, and insects. The team has also validated that acoustic biodiversity data reliably correlates with broader biodiversity, including animals that don't vocalise, making the tool a trusted indicator for whole-ecosystem health. Importantly, the tool is already opening pathways for farmers engaged in regenerative practices to measure and verify land stewardship improvements — a critical step toward legitimate payment for ecosystem services and nature repair markets.

Hear more on the Gulbali Reports Podcast of Professor David Watson

Hear the podcast

Interested in finding out more?

If you're looking for updates and direct access to the Bush Bird Classifier when it becomes available, in hosting acoustic sensors on your property, or attending training workshops on using the BBC, please let us know!

Get involved