Automatic Classification of Animal Vocalizations: Generalized Perceptual Linear Prediction Features with Hidden Markov Models

This manuscript describes a framework for the analysis and classification of animal vocalizations. The framework combines generalized perceptual linear prediction (gPLP) features, which incorporate information about the ability of the species under study to perceive sounds, and a hidden Markov model (HMM) classification system. The effectiveness of the framework is shown by analyzing African elephant and beluga whale vocalizations. The features extracted from the African elephant data are used as input to a supervised classification system and compared to results from traditional statistical tests. The gPLP features extracted from the beluga whale data are used in an unsupervised classification system and the results are compared to labels assigned by experts. The development of a framework from which to build animal vocalization classifiers will provide bioacoustics researchers with a consistent platform to analyze and classify vocalizations.