Temporallyaware algorithms for the classification of anuran sounds
 Published
 Accepted
 Received
 Academic Editor
 Barry Brook
 Subject Areas
 Bioinformatics, Computational Biology, Data Mining and Machine Learning, Climate Change Biology
 Keywords
 Global warming, Sound classification, Data mining, Feature extraction, Machine learning, Habitat monitoring
 Copyright
 © 2018 Luque et al.
 Licence
 This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. For attribution, the original author(s), title, publication source (PeerJ) and either DOI or URL of the article must be cited.
 Cite this article
 2018) Temporallyaware algorithms for the classification of anuran sounds. PeerJ 6:e4732 https://doi.org/10.7717/peerj.4732 (
Abstract
Several authors have shown that the sounds of anurans can be used as an indicator of climate change. Hence, the recording, storage and further processing of a huge number of anuran sounds, distributed over time and space, are required in order to obtain this indicator. Furthermore, it is desirable to have algorithms and tools for the automatic classification of the different classes of sounds. In this paper, six classification methods are proposed, all based on the datamining domain, which strive to take advantage of the temporal character of the sounds. The definition and comparison of these classification methods is undertaken using several approaches. The main conclusions of this paper are that: (i) the sliding window method attained the best results in the experiments presented, and even outperformed the hidden Markov models usually employed in similar applications; (ii) noteworthy overall classification performance has been obtained, which is an especially striking result considering that the sounds analysed were affected by a highly noisy background; (iii) the instance selection for the determination of the sounds in the training dataset offers better results than crossvalidation techniques; and (iv) the temporallyaware classifiers have revealed that they can obtain better performance than their nontemporallyaware counterparts.
Introduction
Sound classification has become a major issue in numerous scientific and technical applications. Many techniques have been proposed to obtain the desired sound labelling: some for general purpose (Hinton et al., 2012) and others for specific applications (Cowling & Sitte, 2003).
Although sounds are inherently represented by a time series of acoustic data, it is common to focus on small fragments of audio signals and attempt to classify them without considering the preceding or subsequent sound sections. For this reason, nontemporallyaware (NTA) methods are also frequently applied (Tzanetakis & Cook, 2002; Wang et al., 2006).
In order to clarify the temporal character of the sound in this paper, our interest lies in the evolution of its lowlevel shortduration frames and not to the sequence of acoustic units commonly used in bioacoustics (Kershenbaum et al., 2016).
The main aim of this paper is to analyse and compare temporallyaware and NTA classifiers and to show that the consideration of temporal information clearly improves classification performance.
Let us indicate that the study presented here could eventually be applied to the study of global warming, since the sounds produced by certain animal species have been revealed as a strong indicator of temperature changes and, therefore, of the existence of climate change. Of particular interest are the results provided by anuransound analysis (Márquez & Bosch, 1995), and hence these kinds of sounds are analysed in this paper.
As a widely distributed taxonomic group, anurans are considered excellent indicators of biodiversity. However, frog populations have been experiencing dramatic declines over the past decade due to habitat loss, climate change, and invasive species (Xie et al., 2017). Therefore, longterm monitoring of frog populations is becoming increasingly important in the optimization of conservation policy.
It is worth noting that the sound production mechanism in ectotherms is strongly influenced by the ambient temperature (Llusia et al., 2013). Hence, the temperature may significantly affect the patterns of calling songs by modifying the beginning, duration, and intensity of calling episodes and, consequently, the anuran reproductive activity. The presence or absence of certain anuran calls in a certain region, and their evolution over time, can therefore be used as an indicator of climate change.
The first step in biological species identification involves the recording of different sounds in their natural environment, where different devices can be used. Processing of the recorded sounds can be performed either locally in real time (Aide et al., 2013), or in a remote centre requiring, in this case, a suitable communication system, usually a wireless sensor network, which generally requires informationcompressing technologies (Diaz et al., 2012).
In previous work (Luque et al., 2016), a NTA method for sound classification has been proposed. According to this procedure, the sound is split up into frames. Every frame is subsequently featured using 18 parameters (also called features). The frame features are then compared to certain frame patterns belonging to known sounds, thereby assigning a class label to each frame. Finally the sound is classified by frame voting, for which up to nine different algorithms have been proposed (Luque et al., 2018; Romero, Luque & Carrasco, 2016). For the determination of the sounds which should be included in the training dataset, instance selection and crossvalidation techniques are considered and compared.
However, sounds are inherently made up of a time series of acoustic data. Therefore, if the temporal information of the frame is added to the classification process, then better classification results should be expected. It must be borne in mind that the goal of classification is to recognize species and, more precisely, their different vocalizations.
The paper is organized as follows: section “Materials and Methods” describes the anuran dataset and presents the methodology employed to compare classifiers, by depicting its general schema, the six (three framebased, three segmentbased) approaches to temporallyaware classification, the classification algorithms considered, and the performance metrics employed. The application for the classification of a set of actual anuran sounds is presented in section “Results,” where the results of the six approaches are compared with each other and also with NTA classifiers.
Materials and Methods
Sound dataset
For testing purposes, actual anuran sounds provided by the National Natural History Museum (Fonozoo.com, 2017) have been employed. The sounds correspond to two species, the Epidalea calamita (natterjack toad) and Alytes obstetricans (common midwife toad), with a total of 868 recordings containing four classes of sounds:
E. calamita; mating call (369 records).
E. calamita; release call (63 records).
A. obstetricans; mating call (419 records).
A. obstetricans; distress call (17 records).
A total of 4,343 s of recording have been analysed, with an average duration of 5 s. A common feature of all the recordings is that they have been taken in their natural habitat, with very significant surrounding noise (wind, water, rain, traffic, voices, etc.), which posed an additional challenge in the classification process.
To perform a supervised classification, certain sounds have to be selected as patterns (to be used in the training phase) and others are employed for validation and testing purposes. A common practice is to split the dataset into several disjoint subsets and apply a crossvalidation technique (four folds have been used in the paper). However, use of these noisy recordings as patterns may lead to a decrease in the classification performance. Hence, several other approaches arise as an alternative to crossvalidation. In our case, recordings with relatively low background noise, which were carefully selected by biologists and sound engineers, have been used as patterns.
This approach, usually called instance or example selection, is recommended in order to increase the rate of learning by focusing attention on informative examples (Blum & Langley, 1997; Raman & Ioerger, 2003; OlveraLópez et al., 2010; Borovicka et al., 2012). In order to determine the frame patterns, the experts listen to the recording of the anuran calls and simultaneously consider the spectrogram and the set of MPEG7 features, and label each frame that they consider may belong to any of the possible classes.
The parameters for every classifier are determined by exclusively using the pattern records (training dataset). The remaining elements in the dataset are then divided into two approximately equal subsets used for validation and testing. The validation dataset is employed to determine the hyperparameters of the classifiers, such as the number r of relevant features, the number w of frames to be considered in sliding window (SW), recurrent sliding window (RSW) and hidden Markov model (HMM)–SW, and the number T of recurrent inputs in recurrent neural networks (RNNs). On the other hand, the testing dataset, which includes none of the patterns nor validation sounds, is employed for the evaluation of the performance of every algorithm. Table 1 summarizes the sound and pattern dataset.
Sound class  Sound records  Pattern records  

Number  Seconds  Number  Seconds (pattern section)  Seconds (total record)  
Ep. cal. mating call  369 (43%)  1,853  4  13.89  20.39 
Ep. cal. release call  63 (7%)  311  3  0.99  14.56 
Al. ob. mating call  419 (48%)  2,096  4  1.09  19.72 
Al. ob. distress call  17 (2%)  83  2  3.30  9.80 
Silence/noise  –  –  –  45.20  – 
Total  868  4,343  13  64.47  64.47 
General description of the classification methodology
The general schema of the proposed procedure, depicted in Fig. 1, is based on the following steps:
The sound is split up into 10 ms frames. This is the frame length recommended by MPEG7 since it is the approximate time period for the opening and closing of the anuran vocal cords.
Every frame uses D MPEG7 features: a vector x in ℝ^{D} (ISO, 2001). The series of S frame vectors [x_{1}, x_{2} … x_{S}] makes up the X matrix of dimension M × D, which features the full sound. Feature extraction using MPEG7 features has been chosen because very good results are shown in their description of sound frames for classification purposes, and these features appear as serious competitors to melfrequency cepstral coefficients (MFCC) features, which are widely used in many applications (HerreraBoyer, Peeters & Dubnov, 2003). MPEG7 and MFCC features show similar classification performance and although MPEG7 extraction could require more computational effort, it enjoys several advantages: it is semantically richer (in the sense that it is easier to intuitively grasp its meaning); it is fully standardized for generalpurpose applications; and it presents better performance when a reduction in the number of features is required.
Temporal information is considered by using one of two approaches:

3.1
Framebased approach: for every frame, a vector y of C additional features is constructed (Liu & Motoda, 1998) by applying a function f to the matrix X of the MPEG7 original features and hence y = f(X). Therefore, every frame is featured using D + C features, that is, a vector z = x ∪ y in ℝ^{D+C}. Certain forms of the function f, for instance, are comprised of statistical measures of X, or the concatenation of the vectors corresponding to stacked frames.

3.2
Segmentbased approach: every series of N sound frames is represented by a model using its N × D features.

Every sound fragment, either in terms of a frame or segment, is classified by using one of two approaches:

4.1
The frame features are compared to frame patterns belonging to known species, thereby assigning a class label to each frame. By means of the feature extraction and construction procedures previously described, each sound frame is characterized by D + C features or, equivalently, by a point in an ℝ^{D+C} space defined by its coordinate vector z. N pattern frames are also available where the ith pattern is also represented by a point in the ℝ^{D+C} space with a coordinate vector π_{i}. Each frame is labelled as belonging to a certain class θ out of a total of M classes. The set of pattern frames can be viewed as a cloud of points in ℝ^{D+C} and can be identified by a matrix Π = [π_{1}, π_{2}, …, π_{N}]′ containing the coordinate vectors of the N points. The subset of points in Π belonging to the class θ is denoted by its matrix Π_{θ}. NTA classifiers perform a certain type of comparison between the frame to be classified (represented by its vector y) and the pattern frames (represented by its matrix Π). This comparison is carried out in the space of the ℝ^{D+C} features and its result is called a supervised classification. A wide and representative set of nonsequential supervised classifiers is considered and these are described below.

4.2
The segment models are compared to segment patterns belonging to known species, and a class label is thereby assigned to each segment.

Finally the sound is classified by means of frame or segment voting.
MPEG7 feature extraction and selection
The task of extracting MPEG7 features from every sound frame is accomplished by using three different processes: spectrogram analysis, linear prediction coding, and harmonicity analysis. Hence, D = 18 features are obtained by following Luque et al. (2016), which is summarized in Table 2.
Feature description  Extracting process  

1  Total power  Spectrogram analysis 
2  Relevant power(power in a certain frequency band)  
3  Power centroid  
4  Spectral dispersion  
5  Spectrum flatness  
6,7,8  Frequency of the formants (×3) (The first three formants are considered)  Linear prediction coding 
9,10,11  Bandwidth of the formants (×3) (The first three formants are considered)  
12  Pitch  
13  Harmonic centroid  
14  Harmonic spectral deviation  
15  Harmonic spectral spread  
16  Harmonic spectral variation  
17  Harmonicity ratio  Harmonicity analysis 
18  Upper limit of harmonicity 
As shown below, the consideration of temporal information associated to the frames usually leads us to significantly increase the number of features required. In order to cope with this drawback, a reduction in the number of the 18 original MPEG7 features is proposed by considering the r most significant features of each frame (leading to a vector in ℝ^{r}). Feature selection procedures are employed to determine the relevanceordered set of features and its optimal size (Guyon et al., 2006).
The feature selection technique used in the paper is based on the Jensen–Shannon divergence (Lin, 1991). It obtains the separability of the sound classes for every feature by applying the following procedure:
Consider the set of the N pattern frames represented by the matrix Π = [π_{1}, π_{2}, …, π_{N}]′. Focus on the subset of elements in Π belonging to the kth class θ_{k}, which is denoted by its matrix Π_{k} and, specifically, in the ith pattern frame π_{i} ∈ Π_{k}. The vector π_{i} contains D elements, one for each feature. The jth feature corresponding to the ith pattern frame is denoted as π_{ji}  π_{i} ∈ Π_{k}. Let us denote φ_{jk} as the set of values of the jth feature in every frame belonging to the kth class (θ_{k}): φ_{jk} = {π_{ji}} ∀i  π_{i} ∈ Π_{k}.
Estimate the probability density function (pdf) f_{jk} of the values in ϕ_{jk}, that is, of the jth feature values for those pattern frames belonging to the kth class.
For the jth feature and every pair of classes u and v, an indication is obtained of how separate the corresponding f_{ju} and f_{jv} pdfs are. For this purpose, the Jensen–Shannon divergence is used, which is given by (1) $${D}_{JS}({f}_{ju},{f}_{jv})=\frac{1}{2}{\displaystyle \underset{\infty}{\overset{\infty}{\int}}{f}_{ju}}{\mathrm{log}}_{2}\frac{2{f}_{ju}}{{f}_{ju}+{f}_{jv}}dx+\frac{1}{2}{\displaystyle \underset{\infty}{\overset{\infty}{\int}}{f}_{jv}}{\mathrm{log}}_{2}\frac{2{f}_{jv}}{{f}_{ju}+{f}_{jv}}dx.$$
Every value of the Jensen–Shannon divergence is transformed in the corresponding distance, which is given by (2) $${d}_{JS}\left({f}_{ju},{f}_{jv}\right)\equiv \sqrt{{D}_{JS}\left({f}_{ju},{f}_{jv}\right)}.$$
For the jth feature, the separability index Ψ_{j} is derived, in accordance with (3) $${\text{\Psi}}_{j}\equiv \sqrt[B]{{\displaystyle \prod _{u=1}^{A1}}{\displaystyle \prod _{v=i+1}^{A}}{d}_{JS}\left({f}_{ju},{f}_{jv}\right)},$$where A is the total number of classes and B is the number of pairs of classes u and v, which is given by (4) $$B=\frac{A\left(A1\right)}{2}.$$
The separability index Ψ_{j} for the jth feature is an indicator of how separate the pdfs are corresponding to each class. The more separate the pdfs are, the more useful (or relevant, or significant) that feature is for classification. Hence, the value of Ψ_{j} is used as an indicator of the relevance of the jth feature.
For comparison purposes, two NTA methods are also considered:
NTA classification based on 18 MPEG7 features (NS18).
NTA classification based on the r most relevant MPEG7 features (NSr).
Feature construction
In order to consider the temporal behaviour of a sound, the frames should not be considered one by one, but the preceding and subsequent frames should also be taken into account, that is, their ordered succession should be considered. Several methods have been proposed in the literature to include this temporality (Dietterich, 2002; Esling & Agon, 2012). A number of these methods can be considered framebased, that is, they still classify frames but now the frames are featured with additional information on the temporal context. Alternatively, other approaches are defined as segmentbased as they do not classify isolated frames but instead classify a series of frames (a segment). First, three framebased approaches are described:

Construction of local interquartile range (LIQR) features (Schaidnagel, Connolly & Laux, 2014). The general idea for this feature construction technique is to use the time axis to construct new temporallyaware features. These techniques are commonly based on the values of the features of the frame without considering their order, which is usually called a bag of features. Average values or other related statistics are usually employed.
In the case of the anuran calls to be classified, the typical croaking of a frog is found, while other calls are similar to the sound of a whistle. The croaking sound is produced by repeatedly opening and closing the vocal cords (roughly every 10 ms, equal to the frame length) leading to a series of frames featured with widely spread values (Fay, 2012). On the other hand, the whistlelike sounds are produced by a continuous air flow showing a narrow spread in feature values. For the incorporation of this information into the classification process, a new set of features is therefore constructed that considers the spread of the extracted feature values and not their average. Furthermore, in order to avoid the influence of outliers, the interquartile range (IQR) is selected instead of the standard deviation.
In the implementation used, first for every frame, a ‘window’ centred on that frame is considered, using the closest neighbouring frames. For every original feature, a new derived feature is constructed. To this end, the values of the original feature for every frame in the window are considered. The IQR of these values is computed, and this value is considered the new derived feature. In this way, the number of constructed features is C = D, and hence up to 2 × D features (a vector in ℝ^{2×D}) are now identifying a frame, where C of these features include temporal information. In this approach, a window size of 10 frames (100 ms) has been used.
SW (Aggarwal, 2007). In this technique, also known as frame stacking or shingling, a short window with w frames, centred on each frame, is considered. An oddnumbered value is usually chosen for the window size, that is, w = 2d + 1, where d is an integer. The class θ_{i} for the ith frame is obtained using a classifying function f_{c}, as follows: (5) $${\theta}_{i}={f}_{C}({x}_{id},\cdots ,{x}_{i1},{x}_{i},{x}_{i+1},\cdots {x}_{i+d}),$$where x_{j} ∈ ℝ^{D} represents the feature vector for the jth frame. The D features describing each frame are now those corresponding to all the frames under the window. Therefore, each frame is featured using w × D features (a vector in ℝ^{w×D}). In this approach, the number of features describing each frame can significantly increase, thereby inflicting a major impact on the computing resources required in the classification process. For this reason, only the r most relevant features have been used by applying the aforementioned feature selection techniques.
RSW (Joshi & Dietterich, 2003). This is a method similar to the SW procedure above, except that the classifier now considers not only the features of the frame under the window, but also their previous classification results. Thus, the class θ_{i} for the ith frame is obtained as follows: (6) $${\theta}_{i}={f}_{C}({\theta}_{id},\cdots ,{\theta}_{i1},{x}_{id},\cdots ,{x}_{i},\cdots {x}_{i+d}).$$
NTA classifiers
Every framebased approach (and also the segmentbased autoregressive integrated moving average (ARIMA) model, described below) relies on an underlying NTA classifier. A broad and representative selection of these classifiers has been used throughout this paper: minimum distance (MinDis) (Wacker & Landgrebe, 1971); maximum likelihood (MaxLik) (Le Cam, 1990); decision tree (DecTr) (Rokach & Maimon, 2008); knearest neighbours (kNN) (Cover & Hart, 1967); support vector machine (SVM) (Vapnik, 1998); logistic regression (LogReg) (Dobson & Barnett, 2008); neural network (Neur) (Du & Swamy, 2013); discriminant function (Discr) (Härdle & Simar, 2012); and Bayesian classifier (Bayes) (Hastie, Tibshirani & Friedman, 2005).
All these classifiers have been prototyped using MATLAB. The minimum distance classifier in its training phase obtains the mean value μ_{jk} for the jth feature belonging to the kth class. In the test phase for the ith frame, the distance d_{ik} between the frame features and the mean value of the kth class θ_{k} is obtained in accordance with the expression (7) $${d}_{ik}=\sqrt{{\displaystyle \sum _{j=1}^{D}}{\left({x}_{ji}{\mu}_{jk}\right)}^{2}},$$where x_{ji} is the value of the jth feature corresponding to the ith frame. The class assigned to the frame is that with the minimum distance.
The maximum likelihood classifier is used under a Gaussian probability distribution with full covariance. The neural network classifier is based on a feedforward neural network with a 10neuron hidden layer and a oneneuron output layer. The remaining methods and classifiers have been coded based on builtin MATLAB functions using their default parameters, which are reflected in Table 3.
Classif.  Training function  Test function  Additional function 

MinDis  –  –  
MaxLik  fitgmdist  mvnpdf  
DecTr  fitctree  predict  
kNN  fitcknn  predict  
SVM  fitcsvm  predict  
LogReg  mnrfit  mnrval  
Neur  feedforwardnet; train  net  
Discr  fitcdiscr  predict  
Bayes  fitNaiveBayes  posterior  
HMM  hmmtrain  hmmdecode  kmeanlbg; disteusq 
ARIMA  vgxset; vgxvarx  NTA classifiers  aicbic 
RNN  layrecnet train  net 
Segment modelling and classification
With respect to segmentbased approaches for the introduction of temporal information, the following methods are proposed:

HMMs (Rabiner, 1989). This is a genuinely temporallyaware classifier which takes every sound frame x_{i} ∈ ℝ^{D} and assigns it with a discrete label (Linde, Buzo & Gray, 1980; Brookes, 2006), thereby obtaining an observation O_{i}, which is an integer number c_{k} (a code) in the range [0, C − 1]. The series of observations is assumed to be produced by an HMM made up of N connected states S, where the S_{a} state emits the observation code c_{k} with an emitting probability E_{ak}, and evolves to the state S_{b} with a transition probability T_{ab}. For the recognition of isolated ‘words’ (anuran calls), with a distinct HMM designed for each class, a left–right model is the most appropriate, and the number of states should roughly correspond to the number of sounds (phonemes) within the call. However, the differences in error rate for values of N that are close to 5 are small. The structure and the value of N have been taken from Rabiner (1989) and they are depicted in Fig. 2. The E and T matrices are obtained for each class θ from their corresponding pattern frames Π_{θ} using the forward–backward algorithm (Baum & Eagon, 1967). Once the HMMs are identified, the algorithm takes the series of observations for the full sound segment to be classified (and not only for a single frame), and estimates the probability of being produced by the HMM of each class. The full sound segment is labelled as belonging to the class with the highest probability. When a sound file has to be classified, three alternatives for the determination of the segment length have been explored:
The full sound file (HMMF).
A segment with the same length as the regions of interest (ROI) mean length (HMM–ROI). The ROIs are the segments of the sound patterns containing a valid sound (no silence or noise).
A segment defined by a SW of a certain length (HMM–SW).
This is the classifier recommended in the MPEG7 standard. In this technique, the r most significant values have been used, where r is a parameter to be chosen from the experimentation. Additionally, the HMM classifiers tested in the paper use a 256code (C = 256) quantization codebook.
RNNs (Parascandolo, Huttunen & Virtanen, 2016). The series of frame features x_{i} is introduced into a neural network with H neurons in its hidden layer, which produces an intermediate output y_{i}. The previous outputs y_{i−1} to y_{i−T} are then introduced as new inputs of the network (Fig. 3). A value of H = 10 neurons in the hidden layer is used throughout the paper.

ARIMA models (Box, Jenkins & Reinsel, 2011). The series of frame features x_{i} ∈ ℝ^{D} is considered the result of the vector ARIMA time series, VARIMA(a,d,b), defined as (8) $${x}_{i}^{(d)}={C}_{0}+{\displaystyle \sum _{k=1}^{a}{A}_{k}{x}_{ik}{}^{(d)}}+{\displaystyle \sum _{k=1}^{b}{B}_{k}{\epsilon}_{ik}}+{\epsilon}_{i},$$where a is the order of the autoregressive model, d is the degree of differencing, and b is the order of the movingaverage model. The coefficient matrices A_{k} and B_{k} have a D × D dimension, and the C_{0} vector, representing the time series mean, has D components. In this case, the number of features describing the sound segment is (a + b) × D^{2}. For the sake of simplicity, the stationarity of time series (d = 0) is assumed. On the other hand, VARMA models can be approximated by equivalent VAR models (b = 0). In this case, the optimum value for the remaining order of the model (a) is obtained using the Akaike information criterion (AIC) (Akaike, 1974), and the A_{k} matrices are estimated using a maximumlikelihood technique (Hevia, 2008). Every sound segment, featured with a × D^{2} parameters, can now be labelled using NTA classifiers.
In order to determine the order of the model (a), first the optimum order for every kth ROI pattern (in the training dataset) is computed using a weighted AR mean order ${\overline{a}}_{k}$, derived as (9) $${\overline{a}}_{k}=\frac{{\displaystyle {\sum}_{i=1}^{{O}_{M}}i}\xb7{\text{AIC}}_{ik}}{{\displaystyle {\sum}_{i=1}^{{O}_{M}}{\text{AIC}}_{ik}}},$$where AIC_{ik} is the AIC value for the kth ROI pattern modelled as a VAR model of order i, and O_{M} is the maximum VAR order considered (O_{M} = 10 is used). The optimum value for the VAR order model is then determined by (10) $$a=\frac{1}{{N}_{\text{ROI}}}{\displaystyle \sum _{i}^{{N}_{\text{ROI}}}}{\overline{a}}_{k},$$where N_{ROI} is the number of ROI segment patterns.
Classification performance metrics
The definition of the proper performance indicators constitutes a key aspect in the evaluation of procedures, and it is difficult to overstate its importance (Sturm, 2014). In order to compare the results obtained for every classifier, several metrics for the performance of a classifier can be defined (Sokolova & Lapalme, 2009), all of which are based on the confusion matrix (Table 4).
Classification class  

Classified as positive  Classified as negative  
Data class  Positive  TP (true positive)  FN (false negative) 
Negative  FP (false positive)  TN (true negative) 
The most relevant metrics and their definitions are shown in Table 5, where they are computed for each class that is considered ‘positive,’ as compared to the remaining classes, which are considered ‘negative.’ Additionally, an average value per class can be defined for each metric.
Metric  Formula  Evaluation focus 

Accuracy  $\text{ACC}=\frac{\text{TP}+\text{TN}}{\text{TP}+\text{TN}+\text{FP}+\text{FN}}$  Overall effectiveness of a classifier 
Precision  $\text{PRC}=\frac{\text{TP}}{\text{TP}+\text{FP}}$  Class agreement of the data labels with the positive labels given by the classifier 
Sensitivity  $\text{SNS}=\frac{\text{TP}}{\text{TP}+\text{FN}}$  Effectiveness of a classifier to identify positive labels. Also called true positive rate (TPR) 
Specificity  $\text{SPC}=\frac{\text{TN}}{\text{TN}+\text{FP}}$  How effectively a classifier identifies negative labels. Also called true negative rate (TNR) 
F_{1} score  ${F}_{1}=2\frac{\text{PRC\xb7SNS}}{\text{PRC+SNS}}$  Combination of precision (PRC) and sensitivity (SNS) in a single metric 
Geometric mean  $\text{GM}=\sqrt{\text{SNS\xb7SPC}}$  Combination of sensitivity (SNS) and specificity (SPC) in a single metric 
Area under (ROC) curve  $\text{AUC}=\underset{0}{\overset{1}{{\displaystyle \int}}}\text{SNS\xb7dSPC}$  Combined metric based on the receiver operating characteristic (ROC) space (Powers, 2011) 
Since, in the dataset, the number of instances in every class remains imbalanced (see Table 1), the use of accuracy or precision as the main performance metric can imply a significant skew (Chawla, 2005). It is therefore preferred to use sensitivity and specificity since these remain unbiased metrics even when the classes are imbalanced (GonzalezAbril et al., 2014, 2017). Therefore, when a single metric is required for the comparison of classifier results (i.e. to identify ‘the best classifier’), the geometric mean or the area under curve (AUC) values are preferred since they combine, in a single metric, the sensitivity and the specificity which both present better behaviour in the presence of imbalanced classes. The AUC is more commonly employed and is the metric used for the selection of the best options and/or classifiers throughout the paper. When only one point is available in the receiver operating characteristic (ROC) space, the value of the AUC is computed as the arithmetic mean of sensitivity and specificity.
Confidence interval of the classification performance metrics
Once the classification performance metrics are obtained, it is good practice to estimate the confidence interval of their values. To undertake this task, a bootstrap analysis is performed (Efron & Tibshirani, 1994). Firstly consider the testing dataset 𝒯 containing S sounds. From this dataset, S samples are then taken with replacement and a new 𝒯_{1} dataset is obtained. Due to the replacement in the sampling process, certain sounds are not contained in 𝒯_{1}, while others are repeated at least once. The classification metrics vector μ_{1} can now be computed for the 𝒯_{1} dataset.
This process is repeated N_{b} times (usually a large number), thereby obtaining datasets ${\mathcal{T}}_{1},{\mathcal{T}}_{2}\cdots {\mathcal{T}}_{{N}_{b}}$ and their corresponding metrics vectors ${\mu}_{1},{\mu}_{2},\cdots {\mu}_{{N}_{b}}$. This set of metrics vectors is employed to estimate the pdf of the metrics vector f(μ) and other related statistics. This procedure is commonly employed to derive the confidence interval of the classification metrics. Therefore, considering the metric μ_{k}, which is the kth metric in the μ vector, and its pdf f_{k}(μ_{k}), the confidence interval of μ_{k}, for a given confidence level γ, is the interval between the values u_{k} and v_{k} such that Pr[u_{k} ≤ μ_{k} ≤ v_{k}] = γ. The value of u_{k} can be estimated as the γ/2 percentile of μ_{k}, and the value v_{k} as the 100 − (γ/2) percentile. Throughout this paper, bootstrap analysis with N_{b} = 1,000 and a confidence level of γ = 95% is used.
Bootstrap analysis can also be employed to estimate the probability that a certain metric outperforms another. For every 𝒯_{j} dataset, the classification methods 1 and 2 are employed and their metric vectors μ_{j1} and μ_{j2} are computed. The difference between these metric vectors is then derived by δ_{j} = μ_{j1} − μ_{j2}. The pdf of the differences vector f(δ) and the continuous density function (cdf), F(δ), can then be computed. Finally, considering the difference δ_{k}, which is the kth metric in the δ vector, and its cdf F_{k}(δ_{k}), the probability of outperforming, o_{k}, is the probability that δ_{k} > 0, that is, o_{k} = Pr[δ_{k} > 0] = F_{k}(0).
Results
Instance selection vs. crossvalidation
In Fig. 4, crossvalidation and instance selection approaches are depicted for NTA classification based on 18 MPEG7 features (NS18). As can be observed, most of the algorithms present a significantly better performance when the patterns are chosen using the instance selection method, with an increase of more than seven points (in %) in the AUC metric of the centroid. Similar results are obtained for other temporallyaware and NTA classifiers for which instance selection has been employed.
NTA classification for a varying number of features
The results obtained by the NTA classifiers based on 18 MPEG7 features are compared using the ROC analysis, which is depicted in Fig. 5. The best result corresponds to the minimum distance classifier, with an AUC of 83.5%. This result is considered as the original baseline (denoted NTA18) for future comparisons.
In order to prevent a high number of features entering the following temporallyaware algorithms, it could be convenient to reduce their number by selecting the r most relevant MPEG7 features. To determine the value of r, the AUC of the validation dataset is used. In Fig. 6, the AUC values for three NTA classifiers are considered as a function of the number of features. The classifiers in the figure are those showing the best AUC performance for values of r: minimum distance, maximum likelihood, and decision tree. From this figure, it can be seen that using the 11 most relevant features (r = 11), the best AUC is obtained. On the other hand, if the computing effort is a major concern and therefore the number of features becomes an important issue, selecting the five most relevant features (r = 5) is a good balance between the AUC and the number of features. A further reduction would produce the steepest AUC decrease (below 75%, decreasing more than 10 points) which is confirmed below (joint optimization subsection). The following subsections derive the results for both the reduced (r = 5) and the optimum (r = 11) number of features.
Classification with a reduced number of features
NTA classification
For comparison purposes, the results obtained by the NTA classifiers based on the five most relevant MPEG7 features are also compared using the ROC analysis, which is depicted in Fig. 7. The best result corresponds to the decisiontree classifier with an AUC of 80.7%. This result is considered as the reduced baseline (denoted NTA5) for future comparisons.
Determining the number of frames
In four of the proposed temporallyaware methods (SW, RSW, HMM–SW and RNN) several consecutive frames have to be considered. The first issue is therefore to determine the optimum number of frames (also called window size w). For this purpose, the AUC of the validation dataset is used, which is represented in Fig. 8 for several temporallyaware methods (using the best underlying NTA classifiers) as a function of the number of frames. In that figure, instead of the AUC absolute value, the increase in the AUC is depicted, compared to the w = 1 case. This graphical approach clearly shows the advantage of using temporallyaware classifiers. In all the methods, except in the RNN, only an odd number of frames have been considered because they are preferred in those algorithms.
From this figure, by using a window size between three and nine in the SW method, the AUC value can be enhanced by more than six points (in %). With these considerations, a sevenframe SW (w = 5) has been selected (its optimum value is denoted in the figure by a filled blue marker). This means a duration of 70 ms which roughly corresponds to seven opening periods of the anuran vocal cords. Similarly, the optimum values of the number of frames for the remaining methods are RSW: 11; HMM–SW: one; and RNN: 15.
LIQR classification
The first framebased approach to temporallyaware classification is now considered: that of the construction of the LIQR features. The results corresponding to the ROC analysis are depicted in Fig. 9. The best result corresponds to the decisiontree classifier that has an AUC of 85.2%. For most of the classifiers, the LIQR approach attains slightly better results than does the equivalent NTA classifier: a mean enhancement of about five points (in %) is achieved in the AUC value compared to the reduced baseline (NTA5).
SW classifiers
By considering the five most relevant features and a sevenframe window size, the SW method (SW75) is examined and its results compared through the ROC analysis, as presented in Fig. 10. The best result corresponds to the decisiontree classifier, with an AUC of 86.7%, which means an enhancement of about six points (in %) compared to the reduced baseline (NTA5), and an enhancement of about three points (in %) compared to the original baseline (NTA18).
The third framebased approach to temporallyaware classification is now considered: the recurrent SW method. The results corresponding to the ROC analysis are depicted in Fig. 11, when five features (r = 5) and an 11frame window size (w = 11) are considered (RSW115). The best result corresponds to the decisiontree classifier, which presents an AUC of 72.7%. For most of the classifiers, the recurrent SW approach obtains worse results than the equivalent NTA classifier, with a mean decrease of about 13 points (in %) in the AUC.
Segmentbased classifiers
The HMM is the first segmentbased approach to the introduction of the temporal information into the classification process. The HMM takes a sound segment and attempts to classify it as a whole, without any framing. The results corresponding to its ROC analysis are depicted in Fig. 12. In this figure, the five most relevant features (r = 5) are considered. The HMM over a segment defined by a SW (HMM–SW) of size w = 1, that is, over a single frame, obtains the best results among the HMM classifiers, with an AUC of 63.2% which, comparatively, is a poor result. Although the HMM is the classifier recommended in the MPEG7 standard, it is clearly superseded by other NTA techniques.
The second segmentbased approach to temporallyaware classification is now considered: the RNN. Using a hidden layer with H = 10 neurons, five features (r = 5), and 15 frames (w = 15), that is, a number of T = 14 previous intermediate outputs, an AUC of 61.0% has been obtained. This result is also depicted in Fig. 12.
Finally, the ARIMA segmentbased approach is considered. As stated before, a vector AR (VAR) simplified model is considered, where the five most relevant features are used (r = 5). The first step is to determine the order of the VAR model (a) using the AIC criterion on the training dataset, as was described in the “Method” section. The results are depicted in Fig. 13, where the AIC values have been normalized to the [0,1] interval. A white point is drawn at every kth row indicating the weighted AR mean order a_{k} for the kth ROI pattern. The optimum value for the VAR order model is represented in the figure with a vertical white line, and has the value a = 3.36. Its closest integer is used as the VAR order model, a = 3.
Once the ARIMA models are determined, their parameters are classified using NTA classifiers and their performances are also compared using the ROC analysis, as illustrated in Fig. 14. The best result corresponds to the decisiontree classifier with an AUC of 62.0%.
Comparing classifiers
Hitherto, partial results have been presented for every temporallyaware method. In order to obtain an overall perspective, a comparison of the six different methods proposed for temporallyaware classifiers is presented in Fig. 15 and Table 6, where the NTA classifiers (original and reduced baselines) are also considered for reasons of contrast (best results are shown in bold).
Method  Features  Frames  Best classifier  ACC  PRC  SNS  SPC  F_{1}  GM  AUC 

NTA18 (original baseline)  18  –  MinDis  84.12  55.37  76.95  90.03  64.40  83.23  83.49 
NTA5 (reduced baseline)  5  –  DecTr  86.82  76.45  72.60  88.77  74.47  80.28  80.68 
LIQR  10 (2 × 5)  –  DecTr  91.88  79.25  77.49  92.94  78.36  84.86  85.22 
SW  35 (7 × 5)  7  DecTr  92.59  74.11  79.07  94.28  76.51  86.34  86.67 
RSW  55 (11 × 5)  11  DecTr  83.41  57.74  58.52  86.96  58.13  71.34  72.74 
HMMF  5  –  –  75.41  44.64  40.62  82.78  42.53  57.99  61.70 
HMM–ROI  5  –  –  71.88  44.87  42.69  75.80  43.75  56.88  59.25 
HMM–SW  5  1  –  72.35  47.08  45.09  81.26  46.06  60.53  63.18 
RNN  75 (15 × 5)  15  –  66.59  47.81  40.01  81.91  43.58  57.27  60.98 
ARIMA  75 (3 × 5^{2})  –  DecTr  80.94  38.75  38.47  85.50  38.61  57.35  61.98 
Note:
Best results are shown in bold.
Additionally, a ROC analysis has also been accomplished for every method and its results are depicted in Fig. 16. From these results, it can be observed that the best performance corresponds to the SW approach (with an underlying decisiontree classifier). It shows the best AUC metric with a value of 86.7%. The SW method also has the best values for almost every performance metric. The only exceptions are the precision and the F_{1} score (which depends on precision) which, although they present the highest values for the LIQR method, also present good figures for the SW method.
Bootstrap analysis
Using bootstrap analysis on the testing dataset, the pdf of the classification performance metrics can be obtained. The results, focusing on the best temporallyaware method (SW) and considering the AUC pdfs for different window sizes, are shown as a colour map in Fig. 17. The colours represent the probability density for every AUC given a certain window size.
Let us now centre on the SW method using the sevenframe optimum window size (as obtained using the validation dataset) and the reduced number of five features. This case, which is denoted as SW75, is now compared to the two NTA baselines: one with the original number of features (NTA18), and the second with the reduced number of features (NTA5). The pdfs for the AUC in these three cases are depicted in Fig. 18.
Not only can Bootstrap analysis offer the confidence interval for every classification performance metric, but it can also, even more importantly, show how much the optimum temporallyaware classification method (sliding window SW75) improves the results above the two NTA baselines (NTA5 and NTA18). The results for a 95% confidence level are shown in Table 7.
Performance improvement  ACC  PRC  SNS  SPC  F_{1}  GM  AUC  

Baseline NTA5  Mean  5.77  −2.33  6.50  5.51  2.06  6.08  6.01 
Conf. Int.  ±2.70  ±10.3  ±12.3  ±2.08  ±9.94  ±7.38  ±6.72  
Baseline NTA18  Mean  8.46  18.73  2.20  4.25  12.12  3.15  3.22 
Conf. Int.  ±2.82  ±6.94  ±11.9  ±1.93  ±7.98  ±7.05  ±6.51 
The AUC improvement over the two mentioned baselines obtained via the SW method for various window sizes is depicted in Fig. 19. In this figure, the 95% confidence interval is also shown.
Classification with the optimum number of features
Separate optimization
It is now time to turn our attention to the cases when the number of features is not such an important issue and it is affordable to use the r = 11 most relevant features. This number was obtained through an optimization procedure presented above.
Now, again, the next issue is to run a second and separate optimization process to determine the optimum number of frames for the methods requiring such a parameter. For this purpose, the AUC on the validation dataset is used, which is represented in Fig. 20 for several temporallyaware methods (using the best underlying NTA classifiers) as a function of the number of frames. In this figure, instead of the AUC absolute value, the increase of the AUC compared to the w = 1 case is depicted.
From this figure, by using a window size between three and nine in the SW method, the AUC value can be enhanced by about three points (in %). With these considerations, a threeframe SW (w = 3) has been selected (its optimum value, denoted in the figure by a filled blue marker). Similarly, the optimum number of frames for the remaining methods are RSW: three; HMM–SW: 11; and RNN: three.
Repeating the analysis of the various temporallyaware methods on the testing dataset, now using 11 features, the results obtained are presented in Fig. 21 and Table 8, where the NTA classifiers (original and optimum baselines) are also considered for reasons of contrast (best results are shown in bold).
Method  Features  Frames  Best classifier  ACC  PRC  SNS  SPC  F_{1}  GM  AUC 

NTA18 (original baseline)  18  –  MinDis  84.12  55.37  76.95  90.03  64.40  83.23  83.49 
NTA11 (reduced baseline)  11  –  MinDis  88.00  77.67  81.03  92.50  79.31  86.58  86.77 
LIQR  22 (2 × 11)  –  DecTr  89.77  75.89  74.59  91.49  75.23  82.61  83.04 
SW  33 (3 × 11)  3  MinDis  90.47  79.30  82.85  93.93  81.03  88.21  88.39 
RSW  33 (3 × 11)  3  MinDis  90.47  79.30  82.85  93.93  81.03  88.21  88.39 
HMMF  11  –  –  78.24  21.77  50.42  85.20  48.13  65.54  67.81 
HMM–ROI  11  –  –  72.59  85.96  48.39  76.03  61.92  60.66  62.21 
HMM–SW  11  11  –  75.29  45.56  37.02  83.82  39.16  55.70  60.42 
RNN  33 (3 × 11)  3  –  71.06  48.69  47.44  84.56  48.06  63.34  66.00 
ARIMA  363 (3 × 11^{2})  –  MinDis  89.29  48.03  47.88  90.81  47.96  65.94  69.34 
Note:
Best results are shown in bold.
Additionally, an ROC analysis has also been accomplished for every method and its results are depicted in Fig. 22. From these results, it can be observed that the best performance corresponds to the SW (and to the RSW) approach (with an underlying minimum distance). It shows the best AUC metric with a value of 88.4%. The SW (and RSW) method also has the best values for every performance metric. Although SW and RSW methods show exactly the same performance metrics on the testing dataset, the SW has been chosen as the best method for two reasons: it offers slightly better AUC on the validation dataset; and it provides better performance for nonoptimum window sizes (see Fig. 20).
Bootstrap analysis can now offer the confidence interval on how much the optimum temporallyaware classification method (sliding window SW311) improves the results above the two NTA baselines (NTA11 and NTA18). The results for a 95% confidence level are shown in Table 9.
Performance improvement  ACC  PRC  SNS  SPC  F_{1}  GM  AUC  

Baseline NTA5  Mean  2.47  1.64  1.85  1.43  1.74  1.66  1.64 
Conf. Int.  ±2.76  ±3.76  ±11.4  ±1.76  ±6.63  ±6.62  ±6.20  
Baseline NTA18  Mean  6.36  23.94  6.06  3.91  16.70  5.08  4.98 
Conf. Int.  ±2.94  ±5.28  ±11.6  ±1.85  ±7.02  ±6.71  ±6.25 
Joint optimization
In the previous section the numbers of features and frames were separately optimized, that is, firstly the optimum number of features was determined and, subsequently, the optimum window size for that value was derived.
However, it is also possible to run a joint optimization process to simultaneously seek the optimum values for both parameters. By running this process on the validation set for the best temporalaware method (SW), a set of AUC values for each pair of values of the parameters (number of features and window size) is obtained. The result is shown in Fig. 23 in the form of several lines (one per window size) that depict the increase of the AUC compared to the w = 1 case. In this figure, the maximum values (optimums in the number of features dimension) are represented by small filled circles. This figure also confirms that the selection of the five most relevant features (r = 5) provides a good balance between the AUC and the number of features (a result previously derived from Fig. 6).
An alternative way to represent the joint optimization process is to employ a bidimensional colour map, as in Fig. 24, which depicts the increase of AUC for every pair of values (number of features, window size). The optimums in the number of features dimension are marked with a black spot, while the optimums in the window size dimension are denoted by an empty circle. The overall optimum value, which is indicated with a cyan filled circle, is reached for eight features and a sevenframe window (SW78).
In order to ascertain the impact of optimizing in each direction, Fig. 25 has been constructed. Given a certain number of features (xcoordinate for the corresponding blue point), the AUC can be optimized by changing the window size, given by a vertical movement in Fig. 24, and the maximum value is the ycoordinate for that blue point. Alternatively, given a certain window size (xcoordinate for the corresponding green point), the AUC can then be optimized through the selection of the proper number of features, given by a horizontal movement in Fig. 24, and the maximum value is the ycoordinate for that green point. It can be seen that, in almost every case, the optimization of the window size offers greater improvement than the optimization of the number of features.
In Fig. 23, the AUC is plotted as a function of the number of extracted (primary) features, which has been denoted as D. However, the SW method adds other C = w·D constructed (secondary or derived) features. Therefore, the total number of features defining the dimension of the space for classification purposes is D + C, and this is the value which has to be considered and kept as low as possible in order to reduce the computing requirements. For this reason, it is worth redrawing this figure in terms of the total number of features. The result is shown in Fig. 26. It can be seen that, if the total number of features is a concern, then the green line (corresponding to w = 3 frames) has a suboptimum (a secondary peak identified with an empty green circle) corresponding to five extracted features, that is, 15 total features. The corresponding classifier (SW35) should also be considered as a possible alternative.
Summary of results
Throughout the previous subsections, several classification methods have been identified. Firstly, there are three NTA classifiers using: the original number of features (NTA18), the reduced or balanced number of features (NTA5) and the optimum number of features (NTA11). These three classifiers have been used as the baselines to determine the improvement achieved using other procedures. Later, when considering temporallyaware methods, the SW method has shown itself to be the most efficient. The determination of the window size in a separate optimization process identifies a classifier for a balanced number of (primary) features (SW75) and another classifier for the optimum number of features (SW311). On the other hand, the joint optimization of the number of features and frames leads to the detection of an optimum method (SW78) or of a classifier that balances the performance metric and the dimension of feature space (SW35). Table 10 summarizes these seven classification methods.
Method  Best classifier  Features  Temporally aware  Optimization  Features concern 

NTA18  MinDis  18  No  No  Original 
NTA5  DecTr  5  No  No  Balanced 
NTA11  MinDis  11  No  Only features  No 
SW75  DecTr  35 (7 × 5)  Yes  Separate  Balanced 
SW311  MinDis  33 (3 × 11)  Yes  Separate  No 
SW35  DecTr  15 (3 × 5)  Yes  Joint  Balanced 
SW78  DecTr  56 (7 × 8)  Yes  Joint  No 
Using bootstrap analysis, the pdf of each performance metric for each classification method can be estimated. The results regarding AUC are shown in Fig. 27 with the classification methods ordered in terms of the increasing number of total features. It can be seen that all the SW classifiers (except the SW35) obtain very similar results and improve the original baseline by about five points (NTA18).
However, by considering another classification performance metrics, different results can be obtained. For instance, Fig. 28 depicts the comparison of the bootstrap analysis when the accuracy (ACC) is considered. Now all the SW classifiers clearly outperform the NTA methods. The best classifier (SW78), which was obtained by a joint optimization process, increases the original baseline (NTA18) by more than 10 points.
In order to compare each classification method using various performance metrics, a box plot has been drawn (Fig. 29). For each metric and each method, four elements are drawn: a filled box from the 25% to 75% percentiles; an upper vertical line from the 75% percentile to the upper limit of the confidence interval; a lower vertical line from the 25% percentile to the lower limit of the confidence interval; and a horizontal black line corresponding to the median value.
This same information is also presented in Table 11 (best results are shown in bold). Using AUC as the single performance metric, the overall best classifier is the SW311 which outperforms the baseline by about five points (requiring 33 features instead of 18). However, the SW78 classifier (requiring 56 features) outperforms SW311 in terms of accuracy, precision, specificity and F_{1} score, and stands as the second best in terms of AUC. On the other hand, if the number of features is the greatest concern, then the NTA11 classifier (requiring 11 features), still outperforms the original baseline with a much reduced number of features.
Method  Features  Statistic  ACC  PRC  SNS  SPC  F_{1}  GM  AUC 

NTA5  5  Mean  2.69  21.06  −4.31  −1.27  10.06  −2.94  −2.79 
NTA11  11  Mean  3.89  22.30  4.21  2.48  14.96  3.42  3.34 
SW35  15 (3 × 5)  Mean  7.05  20.60  −2.13  2.44  10.95  −0.06  0.16 
SW311  33 (3 × 11)  Mean  6.36  23.94  6.06  3.91  16.70  5.08  4.98 
SW75  35 (7 × 5)  Mean  8.46  18.71  2.20  4.25  12.12  3.15  3.22 
SW78  56 (7 × 8)  Mean  9.29  28.55  3.83  4.32  17.87  4.07  4.07 
Note:
Best results are shown in bold.
By considering not only the mean value of the improvements but also their statistical distribution, the confidence interval for each metric and method can be derived. These results are shown in Table 12, where the probability that the chosen method outperforms the original baseline (NTA18) is also presented (best results are shown in bold). It can be seen that for almost every metric, the selected method outperforms the original NTA classifier with a high probability.
Method  Features  Statistic  ACC  PRC  SNS  SPC  F_{1}  GM  AUC 

NTA11  11  Mean  3.89  22.30  4.21  2.48  14.96  3.42  3.34 
Conf. Int.  ±2.94  ±5.12  ±11.5  ±1.90  ±6.94  ±6.72  ±6.29  
Pr. Outperf.  99.42  99.83  77.52  99.38  99.82  85.22  86.23  
SW311  33 (3 × 11)  Mean  6.36  23.94  6.06  3.91  16.70  5.08  4.98 
Conf. Int.  ±2.94  ±5.28  ±11.6  ±1.85  ±7.02  ±6.71  ±6.25  
Pr. Outperf.  100  99.97  86.08  100  99.91  93.81  94.56  
SW78  56 (7 × 8)  Mean  9.29  28.55  3.83  4.32  17.87  4.07  4.07 
Conf. Int.  ±2.71  ±10.3  ±12.0  ±1.91  ±9.01  ±7.07  ±6.54  
Pr. Outperf.  100  100  74.67  99.99  99.86  88.02  89.51 
Note:
Best results are shown in bold.
Discussion
We show that instance selection for the classification of the sounds in the training dataset offers better results than do crossvalidation techniques. This is consistent with several other studies that have shown that selective learning helps reduce the effect of the noise in the data (Raman & Ioerger, 2003; OlveraLópez et al., 2010). As has been addressed in Borovicka et al. (2012), many instances in the training set may prove to be useless for classification purposes and they commonly do not improve the predictive performance of the model and may even degrade it. Despite the noise in the data, certain researchers (Blum & Langley, 1997) have proposed a further two reasons for instance selection. The first reason arises when the learning algorithm is computationally intensive; in this case, if sufficient training data is available, it makes sense to learn from only a limited number of examples for purposes of computational efficiency. Another reason arises when the cost of labelling is high (e.g., when labels must be obtained from experts). In our case, the identification of the first and final frames of the ROIs is a burdensome task which can be minimized by using fewer examples in the training dataset.
Furthermore, from the results, the decisiontree method appears as one of the best classifiers in many temporallyaware methods. This fact is consistent with other studies where nonspeech sounds (Pavlopoulos, Stasis & Loukis, 2004), or more specifically, environmental sounds (Bravo, Berríos & Aide, 2017) are considered.
Additionally, the temporallyaware classifiers have revealed that they can outperform their NTA counterparts. Several authors (Dietrich, Palm & Schwenker, 2003; Salamon et al., 2016) have reached similar results in the field of bioacoustics and argue that the constructed features can better capture spectrotemporal shapes that are representative of the various sound classes.
The SW method attained the best results in our tests, which is also consistent with other works (Salamon & Bello, 2015) that shows that feature learning is more effective when the learning is performed jointly on groups of frames. In their study the authors have reported very similar results for various window sizes. Our study, however, which has comprised a larger set of values for window size, concludes that there is an optimum region for the optimum number of frames and that overly large values of this parameter can even degrade the classification performance. The results attained by the SW method even outperformed the HMM usually employed in speech recognition applications. This result is mainly due to the fact that the HMM is a classifier that uses subword features, which are not suitable for nonspeech sound identification since environmental sounds lack the phonetic structure that speech possesses (Cowling & Sitte, 2003). It has been found that the optimum classifier (SW311) increases the AUC by approximately five points and obtains a noteworthy overall accuracy of 90.5% (six points higher than the baseline). Since the level of background noise in the recordings is high, this can be considered a remarkable result. In Salamon & Bello (2015), an increase of 1.5 points in the AUC and five points in accuracy were reported for an eightframe window size.
The outperformance using these methods may only be moderate but it is reliably consistent. The probability that the selected temporallyaware methods improve their NTA counterparts is extremely high (more than 90% in most cases).
Conversely, the cost of more complex computing due to the higher number of features required in the optimum SW311 method has been considered in detail (Luque et al., 2017). Using 33 (3 × 11) features almost double the number of the original 18 parameters which affects processing efforts in three different aspects. The first issue involves the time required for the construction of the new features that, for the threeframes SW optimum method, is approximately 10 μs measured on a conventional desktop computer. This time is negligible compared to the classification time (detailed below) and to the frame length (10 ms, 1,000 times higher).
Additionally employing a greater number of features leads to higher processing requirements in the task of training classifiers. By doubling the number of features, the time needed to train classifiers is also approximately doubled, with values of 30 ms for the minimum distance and of 800 ms for the decision tree. Although these values are greater than the 10 ms window length they have a limited effect on the overall classification process because classifiers are trained offline only once and, therefore, they do not affect realtime performances.
A third issue regarding processing efforts is the effect of employing a greater number of features on classification times. Using 33 features instead of the original 18 parameters approximately increases the classification time of a frame by 5% with absolute values of approximately 2 ms for the minimum distance and 0.8 ms for the decision tree. Hence, a very limited rise in the computing effort is demanded when the temporallyaware methods are applied.
Another issue to be considered is the ability of the proposed method to identify when within each audio recording the call is located. It could be thought that SW classifiers are going to blur the edges of audio events by using features obtained over a wider time span. However, these classifiers and certain other temporallyaware methods can still sharply identify the events. The SW method features a frame considering preceding and subsequent frames, but it still independently classifies every frame, thereby allowing the precise identification of calls as has been shown in several independent studies (Mesaros, Heittola & Virtanen, 2016; Stowell & Clayton, 2015; Foggia et al., 2015).
Conclusion
Changes in the sounds of anurans can be used as an indicator of climate change. Algorithms and tools for the automatic classification of the different classes of sounds could be developed for this purpose. In this paper, six different classification methods based on the datamining domain have been proposed, which try to take advantage of the temporal behaviour of sound. The definition and comparison of this behaviour is undertaken using several approaches.
A detailed analysis of the classification errors shows that most errors occur when the recordings are very noisy. Additionally, other misclassifications appear when a recording, labelled as belonging to a certain class, is in fact made up of two or more overlapping sounds: one belonging to the true class and the others to a false class.
Firstly, it has been shown that instance selection for the determination of the sounds in the training dataset offers better results than do crossvalidation techniques.
Additionally, the temporallyaware classifiers have revealed that they can obtain a better performance than their NTA counterparts. The SW method attained the best results in our tests, and even outperformed the HMM usually employed in speech recognition applications.
For classifiers based on a given number of features, the optimization of the window size can increase the AUC value by up to 12 points (in %), while the optimization of the number of features only leads to an AUC increase of fewer than three points.
If the number of total features is of no great concern, then the optimum classifier for our dataset is based on 11 original features and a window with three frames (SW311), which increases the AUC by about five points and obtains a noteworthy overall accuracy of 90.5%: a result even more significant when one considers the high level of background noise affecting the sounds under analysis.
On the other hand, if the number of features has to be minimized due to low computing capacity then the optimization of the number of features in NTA classifiers presents the best method, with an optimum for 11 features (NTA11) thereby achieving an increase in the AUC of three points. If a further reduction in the number of features is required, a good compromise is found in the use of only five features (NTA5) instead of the original 18, which reduces the number of parameters to less than one third while it reduces the AUC performance by only three points.