Dictionary Learning for Bioacoustics Monitoring with Applications to Species Classification

Abstract

This paper deals with the application of the convolutive version of dictionary learning to analyze insitu audio recordings for bio-acoustics monitoring. We propose an efficient approach for learning and using a sparse convolutive model to represent a collection of spectrograms. In this approach, we identify repeated bioacoustics patterns, e.g., bird syllables, as words and represent new spectrograms using these words. Moreover, we propose a supervised dictionary learning approach in the multiple-label setting to support multi-label classification of unlabeled spectrograms. Our approach relies on a random projection for reduced computational complexity. As a consequence, the non-negativity requirement on the dictionary words is relaxed. Furthermore, the proposed approach is well-suited for a collection of discontinuous spectrograms. We evaluate our approach on synthetic examples and on two real datasets consisting of multiple birds audio recordings. Bird syllable dictionary learning from a real-world dataset is demonstrated.Additionally, we successfully apply the approach to spectrogram denoising and species classification.

Computational sustainability

“…Other challenges concerning quantification and visualization of uncertainty in species prediction, multiscale data fusion and interpretation from multiple sensors, incorporation of biological and ecological constraints, and models of migration have also been addressed. 30,[32][33][34] Sheldon and collaborators introduced collective graphical models, which can model a variety of aggregate phenomena, even though they were originally motivated for modeling bird migrations 6,32 (Figure 7).…”

Phoneme Classification Using Temporal Tracking of Speech Clusters in Spectro-temporal Domain

“…The main goal of speech features extraction methods is the extraction of valuable discriminative information in the extracted features while reducing the amount of data to a minimum. Mel scaled frequency cepstral coefficients and spectro-temporal features [1][2][3] are the most frequently used representations of the speech signal that are both inspired by the human auditory model. The auditory model is inspired based on psychoacoustical and neurophysiological findings in the human auditory system.…”