Researchers Identify Fish Species Using Sounds

Researchers at the University of Victoria report in the Journal of Fish Biology that they triangulated more than 1,000 underwater sounds in Barkley Sound, B.C., and linked calls to eight rocky-reef species using an acoustic localization array and paired video. A machine-learning model using 47 acoustic features classified species with up to 88% accuracy, potentially enabling noninvasive monitoring and acoustic size estimates for conservation.