Full-stack bioacoustics: field kit to AI to action

The biodiversity crisis is coming into sharp focus. The World Economic Forum ranked biodiversity loss as one of the top 5 global risks, both in impact and likelihood, for the first time in 2020 and again in 2021 (World Economic Forum 2021). Continental-scale assessments warn of population declines in major taxa such as birds and insects (NABCI 2016, Wagner 2021), while also highlighting some major gaps in information (Montgomery 2020). Data for monitoring biodiversity change remain incomplete and (spatially, temporally and taxonomically) biased. It is crucial to monitor the world's wildlife better, and to provide rapid intelligence that can enable us to manage these risks.

Acoustic data (sound recordings) are a vital source of evidence for detecting, counting, and distinguishing wildlife (Brown & Riede 2017, Ganchev 2017, Stowell & Sueur 2020). Sound can be used for rapid and low-cost wildlife monitoring, as well as for research in ecology, animal behaviour, and more. This domain of "bioacoustics" has grown in the past decade due to the massive advances in signal processing and machine learning, the falling costs of recording devices, as well as the affordability and capacity of data processing and storage (Hill et al. 2018). Novel technologies such as bioacoustic AI are needed to maximise opportunities for automating and expanding the extent and resolution of biodiversity monitoring.

There is an information gap, and an opportunity to bridge it. However, there is also an important capability gap. Numerous research papers describe the use of Raspberry Pi or similar devices for acoustic monitoring, and other research papers describe automatic classification of animal sounds by machine learning. But for most ecologists, zoologists, conservationists, the pieces of the puzzle do not come together: the domain is fragmented, meaning that researchers lack the tools or the skills to deploy advanced computational bioacoustics in the field.

In this Lorentz workshop we bridge this gap by bringing together leading exponents of open hardware and open-source software for bioacoustic monitoring and machine learning, as well as ecologists and other field researchers. We share skills while also building a vision for the future development of "bioacoustic AI".