The 10th edition of the World Ecoacoustics Congress was held in Madrid between July 8th and July 12th. In this context, Juan Sebastián Ulloa and myslef have co-organized a special 3-hour symposium titled “Machine listening meets passive acoustic monitoring”. This event is supported by CAPTEO and PETREL projects. We reproduce the symposium abstract below.
Link to symposium page: https://ecoacoustics2024.org
Passive acoustic monitoring (PAM) has the potential to transform ecoacoustic research by providing non-invasive insights into biodiversity and ecosystem dynamics through massive and standardized collection of environmental sound. Currently, a major limitation is the insufficient capability to analyze the vast quantities of recorded sounds. Thus, machine listening has a key role to play in the automation of audio content analysis and the deployment of PAM to large spatiotemporal scales. We define machine listening as pattern recognition in the broadest possible sense: i.e., both for ecoacoustic scenes and events and both in supervised and self-supervised settings. However, there are insufficient datasets for model training and development, as well as a shortage of benchmarks to gauge model efficacy. Furthermore, the absence of coordination and synergy among researchers inhibits the optimization of machine listening for ecoacoustic research. This symposium will bring together researchers from computer science, ecology, and acoustics, so as to map the current state and prospects of how AI is used within ecoacoustics.