This innovative research, led by experts from the University Institute of Marine Research (INMAR) at UCA, addresses a significant challenge in the study of marine fauna. The Strait of Gibraltar, known for its chaotic acoustic environment due to maritime traffic and natural sounds, represents a particularly difficult setting for cetacean monitoring.
Cetaceans fundamentally rely on sound for orientation, communication, and feeding. Therefore, analyzing their vocalizations is vital for understanding their presence and behavior. The AI model developed by Alba Márquez, Neus Pérez, Daniel Benítez, Gonzalo M. Arroyo, and Andrés de la Cruz has been trained to operate effectively in complex acoustic scenarios, overcoming the limitations of traditional systems that require controlled conditions.
This approach, published in the prestigious journal Engineering Applications of Artificial Intelligence, highlights the importance of adapting the model to the acoustic conditions of the environment.
The team collected over 1,300 hours of sound recordings near the island of Tarifa, at a depth of ten meters, between May 2024 and March 2025. During these deployments, the model was validated across different seasons, demonstrating its ability to detect whistles even with high levels of noise and sound overlap.
The system's performance improvement was achieved through progressive adaptation with local data and iterative validation, proving far more effective than conventional models. This breakthrough is crucial for passive cetacean monitoring, a technique that allows these species to be tracked without altering their natural behavior.
The automation of this process will streamline the analysis of large volumes of data, facilitating the surveillance of ecologically sensitive marine areas. With the Strait of Gibraltar being a key point of marine biodiversity, this research opens new avenues for deeper scientific understanding and the development of more effective conservation initiatives.
