Discriminate animal sounds using TESPAR analysis

Abstract

Identification of animal species by their sounds is important for biological research and biodiversity assessment. In this paper we investigate the use of a dedicated TESPAR analysis, which does not use directly the acoustic signal from animals, but cepstral coefficients derived from it (MFCC and T-MFCC), to discriminate between different species. Our experiments shows that TESPAR S-matrices of some cepstral coefficients can be successfully used to discriminate between different animal species.