Machine Learning Meets Ecoacoustics: Challenges and Opportunities in Analyzing Complex Signals from Marine MammalsÂ
The study of sound and video in natural environments, such as marine protected areas, provides valuable insights into marine ecosystems and informs conservation efforts. Our research focuses on developing innovative solutions for passive acoustic and video monitoring (PAVM) and ecoacoustic analysis, leveraging machine learning techniques to analyze complex signals from marine mammals. We discuss the potential benefits of using machine learning approaches to classify orca vocalizations, paving the way for more accurate population census and habitat monitoring. Building on this work, we developed MASE, an instrument designed to record underwater soundscape, which provides real-time ecoacoustic indices. By integrating PAVM analysis and machine learning, we can gain a deeper understanding of marine ecosystems and support evidence-based conservation efforts.