1T-02
visualizing soundscapes and quantifying interspecific interactions in forest animal vocalizations using robot audition technology
This paper aims at visualization and quantification of forest animal vocalizations by applying robot audition techniques and ecoacoustic methods. We create a false-color spectrogram of which colors reflect the directions of arrival of vocalizations of birds and cicadas which are localized and separated by HARK, which visualizes directional changes in their vocalizations. We further propose an automatic classification method to classify their vocalizations by using an acoustic index (acoustic cover, CVR). We conduct a proof-of-concept experiment and analysis of their localized vocal durations, which implies there exist interspecific interactions between birds and cicadas.