Acoustic Identification Research Articles

Outdoor Acoustic Event Identification using Sound Source Separation and Deep Learning with a Quadrotor-Embedded Microphone Array

Outdoor Acoustic Event Identification using Sound Source Separation and Deep Learning with a Quadrotor-Embedded Microphone Array

冬タイヤ用音響同定システムの開発

冬タイヤ用音響同定システムの開発

Is it possible to acoustically identify individuals within a population?

Acoustically identifying individuals may be a helpful technique when it is necessary to monitor animal populations over space and time. Previous studies have largely focused on the theoretical exploitation of vocal individuality or have looked at a small number of individuals. Here, we examined whether vocal individuality can be used to track the movement of individuals within a population (in this case when the number of individuals is greater than 100) and unknown beforehand. As a model species, we used the Corncrake (Crex crex)—a highly secretive bird whose calls are characterized by an individual-specific feature: pulse-to-pulse duration (PPD). When we performed classical discriminant function analyses on PPD, we correctly identified a high percentage of individuals (>98 %), even when sample size was larger than 100. However, a comparison of PPD similarity within and between individuals showed that, while birds can be correctly discriminated, unambiguous identification is impossible when the number of individuals is unknown beforehand. Therefore, we were only able to assess the probability that two calls belonged to the same individual. The results of this study show that acoustic identification in the Corncrake, and probably in other animal species, is mainly useful in detecting general behavioral patterns within populations. For instance, we discovered that more than 50 % of males change territories during the breeding season, probably to find females. Physical marking methods seem to be more reliable to tracking specific individuals. However, those methods usually consider limited numbers of individuals. Therefore, generalizing results to the population scale can also be misleading.

Variation in search-phase calls ofLasiurus varius(Chiroptera: Vespertilionidae) in response to different foraging habitats

Intraspecific variability in echolocation calls could be an important factor hampering the accurate acoustic identification of species in the field. We studied variations in the echolocation behavior of Lasiurus varius in relation to habitat structure. Echolocation calls of L. varius reflected the degree of clutter present in its foraging areas. Bats foraging in an uncluttered habitat emitted longer and lower-bandwidth calls, with the lowest frequency values, whereas bats foraging in a cluttered habitat broadcast shorter and broadband signals, showing an increase in frequency content of their calls. Discriminant function analysis gave an overall classification of 76% of the calls emitted in the different flight situations. Our results highlight the need for independent recordings at each study area or habitat type to circumvent potential echolocation call variations, particularly in fragmented landscapes.

Acoustic environment identification using unsupervised learning

Acoustic environment leaves its characteristic signature in the audio recording captured in it. The acoustic environment signature can be modeled using acoustic reverberations and background noise. Acoustic reverberation depends on the geometry and composition of the recording location. The proposed scheme uses similarity in the estimated acoustic signature for acoustic environment identification (AEI). We describe a parametric model to realize acoustic reverberation, and a statistical framework based on maximum likelihood estimation is used to estimate the model parameters. The density-based clustering is used for automatic AEI using estimated acoustic parameters. Performance of the proposed framework is evaluated for two data sets consisting of hand-clapping and speech recordings made in a diverse set of acoustic environments using three microphones. Impact of the microphone type variation, frequency, and clustering accuracy and efficiency on the performance of the proposed method is investigated. Performance of the proposed method is also compared with the existing state-of-the-art (SoA) for AEI.

Inverse scattering analysis in acoustics via the BEM and the topological-shape sensitivity method

Inverse scattering acoustics find practical applications in the detection and imaging of objects embedded in continuous media as well as in finding the optimum geometric configuration of an object to produce a given radiation performance. This work introduces a boundary element method (BEM) approach for the solution of acoustic identification and optimization problems via a topological-shape sensitivity method. The devised optimization tool takes advantage of the inherent characteristics of BEM to effectively solve the forward and adjoint acoustic problems arising in the topological derivative formulation and to deal with infinite domains. The objectives for the identification and optimization problems are to achieve a prescribed sound pressure at a given region of the problem domain. The locus giving extreme values for the topological derivative indicates the optimum positions for the placement of sound-hard scatterers in order to minimize the cost function. The proposed implementation has the ability to deal with initially empty design spaces as well as with design spaces containing pre-existent scatterers. The capabilities of the method are demonstrated by solving a number of identification and optimization problems.

Research of a new sensor method for predicting arbitrary waves through structural configuration analysis

In this article, based on our understanding of sinusoidal acoustic wave loads identification on the beam foundation acoustic sensor, we extend a new approach to identify arbitrary waves of different amplitudes and waveforms. Since in our new cases study, due to the complexity of moving arbitrary wave loads, the conventional Tikhonov combined with the L -curve method is not an effective way to find the suitable regularization parameter used for wave inverse solutions, we use the Arnoldi–Tikhonov regularization method coupled with the generalized cross validation for seeking regularization parameters; this method proved to be better to find the regularization parameter than the L-curve method. In addition, we study displacement response sensitivities of sensor design parameters, such as geometries of the sensor to optimize the design of the sensor. Meanwhile, we also design a sandwich composite beam sensor to replace the structure with traditional materials, and wave reconstruction results are surprisingly good, even in background noise interference level as high as 20%. Therefore, the performance of the new sensor model is enhanced.

Acoustic Identification of Bats in the Southern Western Ghats, India

Bats play crucial roles in ecosystems, are increasingly used as bio-indicators and are an important component of tropical diversity. Ecological studies and conservation-oriented monitoring of bats in the tropics benefit from published libraries of echolocation calls, which are not readily available for many tropical ecosystems. Here, we present the echolocation calls of 15 species from the Valparai plateau in the Anamalai Hills, southern Western Ghats of India: three rhinolophids (Rhinolophus beddomei, R. rouxii (indorouxii), R. lepidus), one hipposiderid (Hipposideros pomona), nine vespertilionids (Barbastella leucomelas darjelingensis, Hesperoptenus tickelli, Miniopterus fuliginosus, M. pusillus, Myotis horsfieldii, M. montivagus, Pipistrellus ceylonicus, Scotophilus heathii, S. kuhlii), one pteropodid (Rousettus leschenaultii) and one megadermatid (Megaderma spasma). Discriminant function analyses using leave-one-out cross validation classified bats producing calls with a strong constant frequency (CF) component with 100% success and bats producing frequency modulated (FM) calls with 90% success. For five species, we report their echolocation calls for the first time, and we present call frequencies for some species that differ from those published from other parts of the species' ranges. This exemplifies the need for more local call libraries from tropical regions to be collected and published in order to record endemic species and accurately identify species whose calls vary biogeographically.

Study of Moving Sinusoidal Wave Load Across Simple Supported Beam for Sensor Structural Configuration

A continuous structure has several response characteristics that make it a good candidate for a sensor to be used in locating an acoustic source. In this paper, based on a beam structure with simple supports on both ends, the response of the structure to transient sinusoidal wave excitations is examined analytically and also verified by a finite element method (FEM). For sensor configuration on this structure, various interesting parameters such as the aperture of the structure, material properties, and thickness are examined by evaluating their effects on structure displacement responses. Results will be used for acoustic wave identification in the future.

A new synthetic detection technique for trackside acoustic identification of railroad roller bearing defects

The detection of railroad roller bearing defects is of great importance for railway traffic. The acoustic defective bearing detector (ADBD) via a trackside acoustic detection system (TADS) is a popular and reliable tool for trackside identification of railroad bearing incipient defects. However, the defective trackside acoustic signal of bearing is confronted with three challenges: demodulation of acoustic signal, removal of Doppler effect and enhancement of defect frequency. This paper proposes a synthetic detection technique to overcome these difficulties successively. In the technique, the signal envelope is first extracted by a new variable-resolution ridge demodulation (VRRD) method. Then the Doppler effect is removed in the envelope via a dynamic signal resampling method. Finally, the rectified defect frequency is enhanced through the means of logarithmic transformation and piecewise linear detrending. The VRRD technique extracts the envelope from the signal time–frequency distribution (TFD) along the Doppler shift curve of resonance frequency traced by an improved ridge extraction algorithm. The resampling method is carried out on the envelope according to the Doppler shift curve of defect frequency obtained by the same ridge extraction algorithm. The sidebands around defect frequency can be suppressed by the logarithmic transformation and the low-frequency non-linear trend can be removed by the detrending method. The signal amplitude is unified at the same time. Two practical bearing signals with inner-race and outer-race defects separately verify the effectiveness of proposed technique.

Acoustic Emission Source Identification Based on Pattern Recognition Method

A new pattern recognition method based on harmonic wavelet packet (HWPT) and hierarchy support vector machine (H-SVM) is proposed to solve the fatigue damage identification problem of helicopter component. In this approach, HWPT is used to extract the energy feature of acoustic emission (AE) signals on different frequency bands and to reduce the dimensionality of original data features. The H-SVM classifier is used to identify the AE source type. A subset of the experimental data for known AE source type is used to train the H-SVM classifier, the remaining set of data is used to test the H-SVM classifier. Also, the pressure off experiment on specimen of carbon fiber materials is investigated. The results indicate that the proposed approach can implement AE source type identification effectively, and has better performance on computational efficiency and identification accuracy than wavelet packet (WPT) feature extraction and RBF neural network classification. DOI : http://dx.doi.org/10.11591/telkomnika.v12i5.3587

Identification of acoustic wavenumber component of fluctuating surface pressure for flow noise analysis

The importance of flow noise is increasing with the increasing popularity of quiet electric vehicles. In general, flow noise covers a wide frequency range and usually the high frequency flow noise is important for the vehicle quality because of the increased sensitivity of human perception to high frequency noise. For predicting high frequency noise, statistical energy analysis (SEA) is often used in the prediction of acoustic insulation and absorption. Some of SEA modeling for flow noise transmission has been reported in the literature. However, most of the SEA modeling needs CFD results which require huge computation time. In vehicle design, sometimes the time required for improvements does not permit time for development of CFD models. In such situations, easy and fast diagnostic method like energy transmission contribution analysis is required. In this paper, the mechanism of the flow noise transmission is reviewed especially in the case of low speed flow of around 80 km/h. The acoustic wavenumber component of fluctuating surface pressure is dominant source of flow noise transmission for low speed flow. Therefore, the acoustic wavenumber component identification using conventional acoustic SEA model and the wind tunnel test results is presented. The identified acoustic wavenumber component is validated by comparing with the acoustic wavenumber component identified from the wavenumber-frequency spectrum to that computed using a CFD model.

ACOUSTIC CLASSIFICATION OF FRESHWATER FISH SPECIES USING ARTIFICIAL NEURAL NETWORK: EVALUATION OF THE MODEL PERFORMANCE

Hydroacoustic techniques are a valuable tool for the stock assessments of many fish species. Nonetheless, such techniques are limited by problems of species identification. Several methods and techniques have been used in addressing the problem of acoustic identification species and one of them is Artificial Neural Networks (ANNs). In this paper, Back propagation (BP) and Multi Layer Perceptron (MLP) of the Artificial Neural Network were used to classify carp (<em>Cyprinus carpio</em>), tilapia (<em>Oreochromis niloticus</em>), and catfish (<em>Pangasius hypothalmus</em>). Classification was done using a set of descriptors extracted from the acoustic data records, i.e. Volume Back scattering (Sv), Target Strength (TS), Area Back scattering Strength, Skewness, Kurtosis, Depth, Height and Relative altitude. The results showed that the Multi Layer Perceptron approach performed better than the Back propagation. The classification rates was 85.7% with the multi layer perceptron (MLP) compared to 84.8% with back propagation (BP) ANN.

Acoustic identification of four species of bats (Order Chiroptera) in central Chile

Echolocation calls of four species of insectivorous bats of central Chile were recorded and characterized to determine vocal signatures that allow their identification in the field. Pulses of Tadarida brasiliensis were characterized by the highest duration and the lowest values for all frequencies, which do not overlap those of the remaining species. Tadarida emits narrowband, shallow frequency-modulated search calls. All three vespertilionid species studied (Histiotus montanus, Lasiurus varius and Myotis chiloensis) showed similar echolocation design to one another, consisting of a downward frequency modulation at the beginning of the signal followed by a narrowband quasi-constant frequency component; however, their calls differ by their spectral characteristics. Discriminant function analysis of six acoustic parameters (duration, initial frequency, slope frequency modulation, peak frequency, minimal and maximal frequencies) gave an overall classification of 87.4%, suggesting species could be correctly classified based on echolocation calls. Call duration and minimal frequency were the variables most important for species identification.

A New Variable Regularized QR Decomposition-Based Recursive Least M-Estimate Algorithm—Performance Analysis and Acoustic Applications

This paper proposes a new variable regularized QR decompPosition (QRD)-based recursive least M-estimate (VR-QRRLM) adaptive filter and studies its convergence performance and acoustic applications. Firstly, variable L2 regularization is introduced to an efficient QRD-based implementation of the conventional RLM algorithm to reduce its variance and improve the numerical stability. Difference equations describing the convergence behavior of this algorithm in Gaussian inputs and additive contaminated Gaussian noises are derived, from which new expressions for the steady-state excess mean square error (EMSE) are obtained. They suggest that regularization can help to reduce the variance, especially when the input covariance matrix is ill-conditioned due to lacking of excitation, with slightly increased bias. Moreover, the advantage of the M-estimation algorithm over its least squares counterpart is analytically quantified. For white Gaussian inputs, a new formula for selecting the regularization parameter is derived from the MSE analysis, which leads to the proposed VR-QRRLM algorithm. Its application to acoustic path identification and active noise control (ANC) problems is then studied where a new filtered-x (FX) VR-QRRLM ANC algorithm is derived. Moreover, the performance of this new ANC algorithm under impulsive noises and regularization can be characterized by the proposed theoretical analysis. Simulation results show that the VR-QRRLM-based algorithms considerably outperform the traditional algorithms when the input signal level is low or in the presence of impulsive noises and the theoretical predictions are in good agreement with simulation results.

Acoustic identification of buried underwater unexploded ordnance using a numerically trained classifier (L)

Using a finite element-based structural acoustics code, simulations were carried out for the acoustic scattering from an unexploded ordnance rocket buried in the sediment under 3 m of water. The simulation treated 90 rocket burial angles in steps of 2°. The simulations were used to train a generative relevance vector machine (RVM) algorithm for identifying rockets buried at unknown angles in an actual water/sediment environment. The trained RVM algorithm was successfully tested on scattering measurements made in a sediment pool facility for six buried targets including the rocket at 90°, 120°, and 150°, a boulder, a cinderblock, and a cinderblock rolled 45° about its long axis.

A continental‐scale tool for acoustic identification of European bats

Summary Acoustic methods are used increasingly to survey and monitor bat populations. However, the use of acoustic methods at continental scales can be hampered by the lack of standardized and objective methods to identify all species recorded. This makes comparable continent‐wide monitoring difficult, impeding progress towards developing biodiversity indicators, trans‐boundary conservation programmes and monitoring species distribution changes. Here we developed a continental‐scale classifier for acoustic identification of bats, which can be used throughout Europe to ensure objective, consistent and comparable species identifications. We selected 1350 full‐spectrum reference calls from a set of 15 858 calls of 34 European species, from EchoBank, a global echolocation call library. We assessed 24 call parameters to evaluate how well they distinguish between species and used the 12 most useful to train a hierarchy of ensembles of artificial neural networks to distinguish the echolocation calls of these bat species. Calls are first classified to one of five call‐type groups, with a median accuracy of 97·6%. The median species‐level classification accuracy is 83·7%, providing robust classification for most European species, and an estimate of classification error for each species. These classifiers were packaged into an online tool, iBatsID, which is freely available, enabling anyone to classify European calls in an objective and consistent way, allowing standardized acoustic identification across the continent. Synthesis and applications. iBatsID is the first freely available and easily accessible continental‐scale bat call classifier, providing the basis for standardized, continental acoustic bat monitoring in Europe. This method can provide key information to managers and conservation planners on distribution changes and changes in bat species activity through time.

Acoustic Identification of Individuals within Large Avian Populations: A Case Study of the Brownish-Flanked Bush Warbler, South-Central China

Acoustic identification is increasingly being used as a non-invasive method for identifying individuals within avian populations. However, most previous studies have utilized small samples of individuals (<30). The feasibility of using acoustic identification of individuals in larger avian populations has never been seriously tested. In this paper, we assess the feasibility of using distinct acoustic signals to identify individuals in a large avian population (139 colour-banded individuals) of Brownish-flanked Bush Warbler (Cettia fortipes) in the Dongzhai National Nature Reserve, south-central China. Most spectro-temporal variables we measured show greater variation among individuals than within individual. Although there was slight decline in the correct rate of individual identification with increasing sample sizes, the total mean correct rate yielded by discriminant function analysis was satisfactory, with more than 98% of songs correctly recognized to the corresponding individuals. We also found that using a part of randomly selected measured variables was sufficient to obtain a high correct rate of individual identification. We believe that our work will increase confidence in the use of using acoustic recognition techniques for avian population monitoring programs.

Application of Neural Network on Acoustic Signal Identification

The application based on Backpropagation (BP) Algorithm network is conducted on identifying the categories and numbers of mechanical equipments by acoustic signal in battlefield targets. Collected signal was pre-processed and extracted the power spectrum feature of acoustic signal as input vectors of neural networks, then classified by neural networks and pattern recognition theorem. We employ the acoustic signals of six kinds of normal equipments as training samples to train the network. The experiment shows that the ratio of recognition of the acoustic signal processing system based on neural networks proposed is better than the conventional methods.

Short Cf-Fm and Fm-short CF Calls in the Echolocation Behavior ofPteronotus macleayii(Chiroptera: Mormoopidae)

Echolocation signals were recorded from the Macleay's bat (Pteronotus macleayii) while hunting for insects in the field. Search, approach and terminal phases were identified in the foraging activity of the species. During search phase, P. macleayii emitted calls consisting of a short CF component followed by a downward FM sweep (sCF-FM), typical of the small Pteronotus species. In addition, FM-sCF calls were recorded. In search flight some sequences consisted only of sCF-FM calls while others consisted only of FM-sCF calls. Most sequences, however, combined both call designs. Based on this and previous studies we suggest that sCF-FM and FM-sCF calls will characterize the vocal repertoire of each of the small Pteronotus species. We discuss the relevance of our results for acoustic identification of P. macleayii and the other small Pteronotus in the field.