Unsupervised acoustic classification of bird species using hierarchical self-organizing maps

Abstract

In this paper, we propose the application of hierarchical self-organizing maps to the unsupervised acoustic classification of bird species. We describe a series of experiments on the automated categorization of tropical antbirds from their songs. Experimental results showed that accurate classification can be achieved using the proposed model. In addition, we discuss how categorization capabilities could be deployed in sensor arrays. © Springer-Verlag Berlin Heidelberg 2007.