Long-term passive acoustic monitoring can provide important insights on the study of biological choruses, which represent a key component of natural environments. Nowadays, the development of methods for analysis and visualization of large acoustic datasets is an active area of research. In this context, the present paper addresses how the traditional computation of spectrograms and Sound Pressure Levels (SPL) could be used for analyzing large sound datasets. Additionally, a visualization tool named here as SPL-Gram and a method for automatic detection of trends in dawn and dusk choruses are presented. The dataset used as a case study represents 3 months of underwater sound collected in a marine wildlife refuge in southern Brazilian coast. Results reveal events with strong daily periodicity, originated by fish choruses in the frequency band from 0.01–2 kHz, and, in the higher frequencies, reflecting acoustic activity of crustaceans. The reported periodicities show a marked relation with sunrise and sunset through the studied period, thus revealing circadian cycles present in the monitored environment. The proposed methodology is not only easy for implementation, but also proves to be valuable in the description of daily and seasonal patterns of biological choruses in large acoustic datasets. |
