Grant Award View - GA356421

Audio monitoring could play a vital role in demonstrating the enhancement of biodiversity values and has been used to monitor species including birds, mammals, frogs and insects. With the rapid advancement of machine learning models, processing audio recording data has become increasingly efficient and accurate. However, a key problem hampering the uptake of audio data as a tool to monitor biodiversity is the large memory-size of audio files and limited availability of mobile network coverage and electricity across most of Australia. These constraints create major bottlenecks for processing audio data that limit the potential to scale-up this technology and use it to monitor biodiversity. In addition to improving the efficiency of data processing, there is also a need for improved post-processing analysis of audio data to demonstrate the enhancement of biodiversity values through a series of easily interpretable metrics with a high level of statistical certainty.

The aim of this project is to eliminate data processing bottlenecks and improve post-processing analysis of audio monitoring data. We will do this by uniting the skills and experience of ecologists experienced in biodiversity data collection in remote areas and validation of datasets with edge computing specialists who have high-level expertise in data processing and analytics. Together, they will develop a streamlined dataflow process for audio monitoring of biodiversity that can be used in large-scale monitoring programs in remote Australia. The outcome will be a fully operational audio data-processing workflow that will lead to demonstrable end-to-end cost reductions in surveying, processing, and analysis. This audio-data processing workflow will boost Australia’s capacity to evaluate the efficacy of the management interventions required to deliver Nature Repair projects.