Adaptive DCTNet for audio signal classification

Abstract

In this paper, we investigate DCTNet for audio signal classification. Its output feature is related to Cohen's class of time-frequency distributions. We introduce the use of adaptive DCTNet (A-DCTNet) for audio signals feature extraction. The A-DCTNet applies the idea of constant-Q transform, with its center frequencies of filterbanks geometrically spaced. The A-DCTNet is adaptive to different acoustic scales, and it can better capture low frequency acoustic information that is sensitive to human audio perception than features such as Mel-frequency spectral coefficients (MFSC). We use features extracted by the A-DCTNet as input for classifiers. Experimental results show that the A-DCTNet and Recurrent Neural Networks (RNN) achieve state-of-the-art performance in bird song classification rate, and improve artist identification accuracy in music data. They demonstrate A-DCTNet's applicability to signal processing problems.

One-shot learning for acoustic identification of bird species in non-stationary environments

“…Such technology can assist towards complete bio-inventories of the study site and generate data about biodiversity composition within groups of taxa at multiple levels [8], [9]. Acoustic monitoring can provide baseline information about specific groups of acoustically active biota, and to generate an index of biodiversity based on the complexity of calls recorded within a region [10], [11]. In this work the taxonomic group of interest is birds; nonetheless the proposed methodology is easily extensible to other groups (e.g.…”