Intelligent Sampling for Colombian Soundscapes Using an Artificial Neural Network.

Information extracted from environmental sounds has been of great importance to the analysis of ecological complexity in natural ecosystems. However, the study of these sounds does not have a universal protocol for the sampling and reduction of large quantities of data that it produces. This paper proposes to use a neural network to optimize the sampling of soundscapes of three Colombian ecosystems. The neural network is trained to identify meaningful temporal windows for audio recording from previously gathered data. This method simplifies the acoustic complexity analysis.