Automatic acoustic recognition of shad splashing using a smartphone

Monitoring the numbers of shad ( Alosa fallax rhodanensis , Rhodanian twaite shad) at their reproduction sites in the Rhone basin is an important step for measuring inter-annual changes in their population size. Manual counting involves listening to detect the sounds of splashes produced by shad during spawning. This is a costly operation, requiring high resource levels under difficult working conditions. In order to automatically estimate the number of migrating shad in rivers, an acoustic signal analysis method is proposed. It is based on short-term spectral analysis, combined with a Gaussian mixture model. Implemented on a smartphone, the application provides a number of advantages, such as mobility, audio recording, spawning detection and counting, and means of communication. The results obtained are very promising, and the deployment of such a device is expected to be of great help for counting shads and locating their spawning sites.