This project aimed to develop tools to track climate impact from sound recordings alone.
Made recordings at 12 locations in the McDowell Sonoran Preserve in collaboration with the McDowell Sonoran Conservancy and the Phoenix Zoo, for a period of 12 months. These recorders were custom built and serviced by a team of stewards from the McDowell Sonoran Conservancy, Parsons Field Institute.
Findings and Impact
A machine-learning model was built that correlated weather variables to psychoacoustic parameters in the sound recordings. An acoustic diversity measure, based on articulation entropy and represents how diverse the sound signals are in the place of the recording. The results show that there is a positive, statistically significant relationship between the acoustic diversity measure and cloud cover, wind speed, and temperature; and an inverse statistically significant relationship between acoustic diversity and dewpoint and visibility. The model was used to predict short-term impact on the environment.