Temporally-aware algorithms for the classification of anuran sounds

Using deep learning for acoustic event classification: The case of natural disasters

This study proposes a sound classification model for natural disasters. Deep learning techniques, a convolutional neural network (CNN) and long short-term memory (LSTM), were used to train two individual classifiers. The study was conducted using a dataset acquired online and truncated at 0.1 s to obtain a total of 12 937 sound segments. The result indicated that acoustic signals are effective for classifying natural disasters using machine learning techniques. The classifiers serve as an alternative effective approach to disaster classification. The CNN model obtained a classification accuracy of 99.96%, whereas the LSTM obtained an accuracy of 99.90%. The misclassification rates obtained in this study for the CNN and LSTM classifiers (i.e., 0.4% and 0.1%, respectively) suggest less classification errors when compared to existing studies. Future studies may investigate how to implement such classifiers for the early detection of natural disasters in real time.

Exploiting the Symmetry of Integral Transforms for Featuring Anuran Calls

The application of machine learning techniques to sound signals requires the previous characterization of said signals. In many cases, their description is made using cepstral coefficients that represent the sound spectra. In this paper, the performance in obtaining cepstral coefficients by two integral transforms, Discrete Fourier Transform (DFT) and Discrete Cosine Transform (DCT), are compared in the context of processing anuran calls. Due to the symmetry of sound spectra, it is shown that DCT clearly outperforms DFT, and decreases the error representing the spectrum by more than 30%. Additionally, it is demonstrated that DCT-based cepstral coefficients are less correlated than their DFT-based counterparts, which leads to a significant advantage for DCT-based cepstral coefficients if these features are later used in classification algorithms. Since the DCT superiority is based on the symmetry of sound spectra and not on any intrinsic advantage of the algorithm, the conclusions of this research can definitely be extrapolated to include any sound signal.

Improving Classification Algorithms by Considering Score Series in Wireless Acoustic Sensor Networks

The reduction in size, power consumption and price of many sensor devices has enabled the deployment of many sensor networks that can be used to monitor and control several aspects of various habitats. More specifically, the analysis of sounds has attracted a huge interest in urban and wildlife environments where the classification of the different signals has become a major issue. Various algorithms have been described for this purpose, a number of which frame the sound and classify these frames, while others take advantage of the sequential information embedded in a sound signal. In the paper, a new algorithm is proposed that, while maintaining the frame-classification advantages, adds a new phase that considers and classifies the score series derived after frame labelling. These score series are represented using cepstral coefficients and classified using standard machine-learning classifiers. The proposed algorithm has been applied to a dataset of anuran calls and its results compared to the performance obtained in previous experiments on sensor networks. The main outcome of our research is that the consideration of score series strongly outperforms other algorithms and attains outstanding performance despite the noisy background commonly encountered in this kind of application.

Optimal Representation of Anuran Call Spectrum in Environmental Monitoring Systems Using Wireless Sensor Networks

The analysis and classification of the sounds produced by certain animal species, notably anurans, have revealed these amphibians to be a potentially strong indicator of temperature fluctuations and therefore of the existence of climate change. Environmental monitoring systems using Wireless Sensor Networks are therefore of interest to obtain indicators of global warming. For the automatic classification of the sounds recorded on such systems, the proper representation of the sound spectrum is essential since it contains the information required for cataloguing anuran calls. The present paper focuses on this process of feature extraction by exploring three alternatives: the standardized MPEG-7, the Filter Bank Energy (FBE), and the Mel Frequency Cepstral Coefficients (MFCC). Moreover, various values for every option in the extraction of spectrum features have been considered. Throughout the paper, it is shown that representing the frame spectrum with pure FBE offers slightly worse results than using the MPEG-7 features. This performance can easily be increased, however, by rescaling the FBE in a double dimension: vertically, by taking the logarithm of the energies; and, horizontally, by applying mel scaling in the filter banks. On the other hand, representing the spectrum in the cepstral domain, as in MFCC, has shown additional marginal improvements in classification performance.