M. B. Manser

Same data, different results? Machine learning approaches in bioacoustics

The paper evaluates machine learning methods for caller identity classification in mammalian vocalizations, finding Mel-frequency cepstral coefficients and random forests provide consistent results across datasets.

A practical guide for generating unsupervised, spectrogram-based latent space representations of animal vocalizations

This paper provides a guide for using spectrogram-based dimensionality reduction to create latent space representations of animal vocalizations, demonstrating its application on meerkat calls.