Microphone Arrays

The Impact Planar Nearfield Acoustic Holography implements the inverse method NAH on the basis of the acoustic impulse response field. It is well adapted to modal analysis and can be used in non anechoic environment. Compared to classical modal analysis methods, IPNAH has some interesting advantages, especially in case of fragile structures like ancient musical instruments or for parametric studies. It is indeed a non intrusive method that permits, thanks to an important number of simultaneous measurement points, an analysis with shorter time and lower number of impulse shocks. Inconvenient of the method are due to the inverse calculation process and hypothesis that induce a lack of precision in the vibration field reconstruction in edges areas, especially when edges boundary conditions are free. Those inconvenient are minor compared to advantages. The principle of this method is presented on the first part of the present paper. A short presentation of some recent applications follows.

The noise of motor vehicles is one of the most important problems as regards to pollution on main roads. However, this unpleasant characteristic could be used to determine vehicle speed by external observers. Building on this idea, the present study investigates the capabilities of a microphone array system to identify the position and velocity of a vehicle travelling on a previously established route. Such linear microphone array has been formed by a reduced number of microphones working at medium frequencies as compared to industrial microphone arrays built for location purposes, and operates with a processing algorithm that ultimately identifies the noise source location and reduces the error in velocity estimation

To increase reliability and safety, industrial plant equipment such as compressors, electric motors, gear trains, and so forth are regularly monitored for damage. A traditional approach for monitoring is for a trained technician to make vibration measurements of the equipment. Inspection of the vibration measurements and their comparison to known healthy/damaged data sets allow for assessing the health status of the machines. This process is repeated at regular intervals. However, it is time-consuming and labor-intensive. It would be greatly convenient both in terms of time and cost to develop a remote acoustic based system to detect the health status of industrial equipment. A microphone phased array machine health monitoring system is proposed to remotely identify and classify machine faults. Acoustic spectral signature analysis has existed for many years, and it is similar to vibration analysis using accelerometers with the advantage that nothing must be mounted on the machine. In addition, acoustic imaging systems using microphone phased arrays have also been in existence for many years. The current state-of-the-art, however, requires an expert human operator to interpret the data from these devices, and therefore it is not suitable for automated, online monitoring. The implementation of a microphone phased array approach to monitor equipment faults in a typical industrial environment presents some unique challenges. A typical industrial plant is a highly acoustic reverberant environment that will result in significant reflections and background noise levels. To be effective, the array will need to be able to monitor multiple machines through the plant, simultaneously. To investigate the potential of the proposed approach, a numerical model of the microphone phased array in a large highly reverberant room was developed. The model was then used to investigate several array designs, study the effect of reflections and reverberation, determine the capability of the system to monitor multiple machines and so forth. These numerical studies revealed that the critical concern is that the array signal to noise ratio must be larger than the noise difference between the loudest and quietest machines being monitored. It was also found that the reverberation of typical industrial plants is not important if a sufficient number of microphones is used in the array, e.g. for a plant with an average absorption coefficient of 10% a minimum of 100 microphones is required in the array.

Sound source localization (SSL) for real-world application demands on-field experiments. The effect of sampling frequency, spacing between microphones, total duration for which the data is recorded, different microphone array configuration, and microphone position are important for the estimation of position of a sound source. In order to realize these aspects, laboratory experiments were carried out to ascertain the limits of these criteria. Cross-correlation (CC) and generalized cross-correlation (GCC) methods are common methods for estimation of time delay between pairs of microphones. The performance of cross-correlation method significantly decreased inside a reverberant indoor space. Estimation of the time delay between two received signals was carried out by using three and four microphones of different configurations. From the estimated time delay, fixed stationary acoustic source localization was done by solving a nonlinear hyperbolic range difference equation. Experimental...

This paper proposes an efficient method based on the steered-response power (SRP) technique for sound source localization using microphone arrays: the volumetric SRP (V-SRP). As compared to the SRP, by deploying a sparser volumetric grid, the V-SRP achieves a significant reduction of the computational complexity without sacrificing the accuracy of the location estimates. By appending a fine search step to the V-SRP, its refined version (RV-SRP) improves on the compromise between complexity and accuracy. Experiments conducted in both simulated-and real-data scenarios demonstrate the benefits of the proposed approaches. Specifically, the RV-SRP is shown to outperform the SRP in accuracy at a computational cost of about ten times lower.

A particle-velocity sensor measures the acoustic particle velocity, instead of the acoustic pressure. A biaxial velocity-sensor (a.k.a. a "u-u probe") consists of two uniaxial velocity sensors, oriented orthogonally and (possibly) displaced in space. This paper investigates what orientations of the two axes would allow azimuth-elevation direction finding, what closed-form formulas to achieve this direction finding via eigen-based parameter estimation algorithms, and what the corresponding Cramér-Rao bounds would be.

A "figure-8" sensor is so labeled because its spatial pattern resembles the character "8" with regard to the sensor's axis. This figure-8 pattern narrows as the sensor's order increases. Using two such figure-8 directional sensors of higher order, oriented perpendicularly to each other - this paper pioneers closed-form signal-processing algorithms to estimate an incident signal's azimuth-elevation bivariate direction-of-arrival. Monte Carlo simulations verify these proposed algorithms' efficacy and statistical closeness to the corresponding Cramér-Rao bounds.

Electromagnetic articulography (EMA) is one of the instrumental phonetic research methods used for recording and assessing articulatory movements. Usually, articulographic data are analysed together with standard audio recordings. This paper, however, demonstrates how coupling the articulograph with devices providing other types of information may be used in more advanced speech research. A novel measurement system is presented that consists of the AG 500 electromagnetic articulograph, a 16-channel microphone array with a dedicated audio recorder and a video module consisting of 3 high-speed cameras. It is argued that synchronization of all these devices allows for comparative analyses of results obtained with the three components. To complement the description of the system, the article presents innovative data analysis techniques developed by the authors as well as preliminary results of the system’s accuracy.

The ability of robots to listen to several things at once with their own “ears”, that is, robot audition, is an important factor in improving interaction and symbiosis between humans and robots. The critical issue in robot audition is real-time processing and robustness against noisy environments with high flexibility to support various kinds of robots and hardware configurations. This paper first overviews activities and issues related to robot audition. Then, it presents the “HARK” robot audition software, which provides three primary functions for robot audition, sound source localization, sound source separation, and separated sound recognition, and then reports their performance. Finally, it discusses future directions in new promising areas as well as robotics.

Accurate estimation of Time-Difference of Arrivals (TDOAs) is necessary to perform accurate sound source localization. The problem has traditionally been solved by using methods such as Generalized Cross-Correlation, which uses the entire signal to accurately estimate TDOAs. However, this could pose a problem in distributed sensor networks in which the amount of data that can be transmitted from each sensor to a fusion center is limited, such as in underwater scenarios or other challenging environments. Inspired by approaches from computer vision, in this paper we identify Scale-Invariant Feature Transform (SIFT) keypoints in the signal spectrogram. We perform crosscorrelation on the signal using only the information available at those extracted keypoints. We test our algorithm in scenarios featuring different noise and reverberation conditions, and using different speech signals and source locations. We show that our algorithm can estimate Time-Difference of Arrivals (TDOAs) and the source location within an acceptable error range at a compression ratio of 40 : 1.

Accurate estimation of Time-Difference of Arrivals (TDOAs) is necessary to perform accurate sound source localization. The problem has traditionally been solved by using methods such as Generalized Cross-Correlation, which uses the entire signal to accurately estimate TDOAs. However, this could pose a problem in distributed sensor networks in which the amount of data that can be transmitted from each sensor to a fusion center is limited, such as in underwater scenarios or other challenging environments. Inspired by approaches from computer vision, in this paper we identify Scale-Invariant Feature Transform (SIFT) keypoints in the signal spectrogram. We perform crosscorrelation on the signal using only the information available at those extracted keypoints. We test our algorithm in scenarios featuring different noise and reverberation conditions, and using different speech signals and source locations. We show that our algorithm can estimate Time-Difference of Arrivals (TDOAs) and the source location within an acceptable error range at a compression ratio of 40 : 1.

Convolutional Neural Networks (CNNs) are a popular choice for estimating Direction of Arrival (DoA) without explicitly estimating delays between multiple microphones. The CNN method first optimises unknown filter weights (of a CNN) by using observations and ground-truth directional information. This trained CNN is then used to predict incident directions given test observations. Most existing methods train using spectrallyflat random signals and test using speech. In this paper, which focuses on single source DoA estimation, we find that training with speech or music signals produces a relative improvement in DoA accuracy for a variety of audio classes across 16 acoustic conditions and 9 DoAs, amounting to an average improvement of around 17% and 19% respectively when compared to training with spectrally flat random signals. This improvement is also observed in scenarios in which the speech and music signals are synthesised using, for example, a Generative Adversarial Network (GAN). When the acoustic environments during test and training are similar and reverberant, training a CNN with speech outperforms Generalized Cross Correlation (GCC) methods by about 125%. When the test conditions are different, a CNN performs comparably. This paper takes a step towards answering open questions in the literature regarding the nature of the signals used during training, as well as the amount of data required for estimating DoA using CNNs.

Performance of the basic delay and sum beamforming is influenced by the reflections of sound from surfaces in the environment. In closed spaces this effect is even more significant, where signals from the high number of close secondary mirror image sources interfere with the primary signal. However, one can also make use of the presence of these image sources. If the geometry of the surrounding surfaces is known, one can also focus on the closest secondary image sources, create secondary sound maps (images) and combine them with the primary ones. The advantage of focusing on the secondary sources is that they are seen under independent direction characteristics, which allows the elimination of real interferences, obstructing objects. In this paper we present the results of our investigations of employing room acoustical considerations to increase efficiency and performance of a random-placed microphone array used in a closed enclosure. A simple case of a shoebox shaped room is used ...

Phased microphone arrays have become a well-established tool for performing aeroacoustic measurements in wind tunnels (both open-jet and closed-section), flying aircraft, and engine test beds. This paper provides a review of the most wellknown and state-of-the-art acoustic imaging methods and recommendations on when to use them. Several exemplary results showing the performance of most methods in aeroacoustic applications are included. This manuscript provides a general introduction to aeroacoustic measurements for non-experienced microphone-array users as well as a broad overview for general aeroacoustic experts.

SODIX is a method for the localization of sound sources including the quantification of their directivities. Other source localization methods commonly assume point monopole sources with uniform directivity. The performance of SODIX is compared in this paper with previously published results [9] of conventional delay-and-sum beamforming and the advanced deconvolution methods DAMAS, DAMAS2, CLEAN-SC, TIDY, and LPD by using the same set of array data. The results show that the resolution of SODIX is only slightly inferior to those of the best advanced beamforming methods.

Due to the widespread use of acoustic arrays, optimisation techniques for array design, focused on improving array performance, have been widely published. This paper exploits the statistical relation between different measures of sidelobe levels and the spacing of elements in random linear arrays made up of a small number of sensors. This paper defines the methodology to obtain maximum probability functions, associating array geometry and performance. These maximum probability functions allow a pre-selection of those array geometries that are more likely to be associated to specified sidelobe level values. This pre-selection results in a significantly reduced computational burden.

During the last decades, vibration analysis has been used to evaluate condition monitoring and fault diagnosis of complex mechanical systems. The problem associated with these analysis methods is that the employed sensors must be in contact with the vibrant surfaces. To avoid this problem, the current trend is the analysis of the noise, or the acoustic signals, which are directly related with the vibrations, to evaluate condition monitoring and/or fault diagnosis of mechanical systems. Both, acoustic and vibration signals, obtained from a system can reveal information related with its operation conditions. Using arrays formed by digital MEMS microphones, which employ acquisition/processing systems based on FPGA, allows creating systems with a high number of sensors paying a reduced cost. This work studies the feasibility of the use of acoustic images, obtained by an array with 64 MEMS microphones (8x8) in a hemianechoic chamber, to detect, characterize and, eventually, identify fail...

This paper presents a novel bird monitoring and recognition system in noisy environments. The project objective is to avoid bird strikes to aircraft. First, a cost-effective microphone dish concept (microphone array with many concentric rings) is presented that can provide directional and accurate acquisition of bird sounds and can simultaneously pick up bird sounds from different directions. Second, direction-of-arrival (DOA) and beamforming algorithms have been developed for the circular array. Third, an efficient recognition algorithm is proposed which uses Gaussian mixture models (GMMs). The overall system is suitable for monitoring and recognition for a large number of birds. Fourth, a hardware prototype has been built and initial experiments demonstrated that the array can acquire and classify birds accurately.

This paper proposes an efficient method based on the steered-response power (SRP) technique for sound source localization using microphone arrays: the volumetric SRP (VSRP). As compared to the SRP, by deploying a sparser volumetric grid, the V-SRP achieves a significant reduction of the computational complexity without sacrificing the accuracy of the location estimates. By appending a fine search step to the VSRP, its refined version (RV-SRP) improves on the compromise between complexity and accuracy. Experiments conducted in both simulatedand real-data scenarios demonstrate the benefits of the proposed approaches. Specifically, the RV-SRP is shown to outperform the SRP in accuracy at a computational cost of about ten times lower.

The wide availability of mobile devices with embedded microphones opens up opportunities for new applications based on acoustic sensor localization (ASL). Among them, this paper highlights mobile device self-localization relying exclusively on acoustic signals, but with previous knowledge of reference signals and source positions. The problem of finding the sensor position is stated as a function of estimated times-of-flight (TOFs) or time-differences-of-flight (TDOFs) from the sound sources to the target microphone, and the main practical issues involved in TOF estimation are discussed. Least-squares ASL solutions are introduced, followed by other strategies inspired by sound source localization solutions: steered-response power, which improves localization accuracy, and a new region-based search, which alleviates complexity. A set of complementary techniques for further improvement of TOF/TDOF estimates are reviewed: sliding windows, matching pursuit, and TOF selection. The paper ...

A hybrid approach to predict the far-field noise generated by an airplane nose landing gear is presented in this paper. The approach consists of a Lattice-Boltzmann Method (LBM) for the calculation of the flow-field around the fully detailed geometry of the landing gear which provides the input for a Ffowcs Williams-Hawkings (FW-H) solver to calculate the far-field noise. Both parts have been validated independently. The method is applied to the nose landing gear of a Gulfstream G550 business jet, for comparison with experimental results obtained in the University of Florida UFAFF wind tunnel. The near-field flow simulation using the LBM method showed good correlation with the PIV measurements of the flow field as well as surface microphone measurements, up to frequencies of about 4kHz. The comparison to experimental far-field results shows good agreement in the midfrequency range of 1-3kHz. At both low and high frequencies the simulations underpredict the measured results more strongly than the near field and surface measurements would suggest, which may be due to experimental limitations. A comparison between the solid and porous formulation of the FW-H solver shows that both method provide nearly equivalent results. However, inclusion of additional surfaces such as part of the fuselage is critical to achieving good results with the solid formulation. The effects of resolution of the near-field simulations are also investigated and show the expected lower cutoff frequency for lower resolutions for both the near-field and the far-field. No differences between the cases with different resolutions are observed up to 1 kHz in the near-field and up 2-3kHz in the farfield.

This study presents an implementation of the field programmable gate array (FPGA) and performance analysis of the multiple signal classification (MUSIC) algorithm for the direction-of-arrival (DOA) estimation of the different sound sources. DOA offers a variety of applications in the field of signal detection, communications, navigation and speech processing. MUSIC is one of the high-resolution DOA estimation techniques and the most promising technique for actual hardware implementation. A microphone array consisting of four equally-spaced microphones (sensors) is used as a basis for this study. As for wideband application, the number of sources should be less than the number of sensors thus a maximum of three sound sources are only considered and their locations are estimated. Four DOAs are used for single source, two sets of DOA for double-source and one set for triple source. Matlab simulations and test bench results are presented showing the effectiveness of the algorithm. All simulations and tests are done using five samples only because of the large logic utilization of the codes when comes to the FPGA implementation. This study utilizes an Altera Stratix II FPGA board for the realization of the MUSIC algorithm in hardware implementation and Verilog HDL serves as the programming language.

A system that implements the nearfield acoustical holography (NAH) technique in air has been designed and placed into operation at the CBS Technology Center. The system utilizes a scanning microphone that may be positioned within a 3.7×3.7×1.2 m volume by a computer controlled robot. Sound pressure data from the scanning microphone and a reference are stored in a mini computer and subsequently processed using nearfield holographic techniques. The resulting reconstructions of the pressure, velocity, and vector intensity fields may be displayed using either a 3-D video graphics system or a digital plotter. A detailed discussion of this NAH system will be presented and reconstructions of the sound field generated by a simple source will be shown.

The control of directivity represents a new stake in the reproduction of sound sources by an electroacoustic device. This paper first reminds the principle of a general method for reproducing the acoustical field radiated by a source, using a multichannel device controlled by digital signal processing. It approximates a given directivity pattern by linear combination of the fields radiated by a set of loudspeakers laid out on the faces of a polyhedron. The present study, derived from previous work, is devoted to the bandwidth improvement and to the optimization of the directivity control. It rests on the preliminary constitution of a set of basic directivities, namely the first spherical harmonics, from which it is possible to tune the cardioicity and the three-dimensional orientation of the synthesized directivity pattern. For the simulation of musical instruments, it is important to maintain independent control on the spatial and spectral characteristics and to put focus on a faithful reproduction of the power spectrum. For that purpose, an analysis/synthesis method of the power spectrum is used to provide the automatic design of a correction filter to be applied on near field recording of the instrument before feeding the source. The results are discussed considering room acoustics criteria and examples of musical instruments.

An electric network frequency (ENF) signal can be found in multimedia recordings due to propagation from the power grid. There a variety of applications based on ENF signal use, such as video and audio stream synchronization, and origin determination of multimedia recordings. In this paper, we propose the use of ENF to synchronize real-time audio streams from different, non-synchronized sound devices, for instance from wireless sound cards or low cost USB sound cards. Synchronization of separate audio streams can be achieved by aligning their embedded ENF signals. Our goal is to find out how accurate a synchronization using ENF at different SNR levels can be. We show simulation and real-time experimental results of audio stream synchronization. We found that with sufficient ENF level we can achieve an accuracy of the estimated delay difference of four samples at 44.1 kHz sampling rate.

A linear array having constant beamwidth over an octave range can be constructed from a special arrangement of element positions and filters which alter the effective aperture with frequency. However, sidelobe level control is complicated by the appearance of an anomalous lobe when the array is steered away from broadside.

This paper deals with cogeneration unit noise measurement by the acoustic camera. Noise is not only measured as the final number of sound power levels, but also its original location is determined with the use of the beamforming algorithm. The properties of the used microphone array are considered and numerically calculated as every different microphone array layout will measure with another resolution. From the frequency spectrum, the possible technical source is determined. The results of noise source visualization show the cogeneration unit case noise decreasing effect while also offering the possibilities for design improvements.

This article analyses and mathematically describes a method for mutual identification of multiple mobile robots. The method interchanges identification numbers among multiple robots via audible beeps multiplexed by frequency. The identification sound is sent only when robots approach to each other. The algorithm was primary designed for e-puck robots, but it can be used in any multi-agent robotic system; each mobile robot has to be equipped with three microphones for determination of the sound's direction, speaker and signal processing microcontroller.

This paper describes a method to record large musical ensembles for use in computer simulations and auralization. The method uses a multimicrophone hemispherical array distributed over an orchestra playing on a hard floor. A ray tracing program models the orchestra as sectors of a hemisphere corresponding to each microphone location. Microphone signals are then mapped to their respective source sectors by convolution with the model's outputs. The method permits recording and simulating an entire orchestra with its directivity, including the movement of the musicians while playing. The approach is geared towards achieving a high quality of orchestral musicianship and sense of orchestral ensemble while maintaining high spatial accuracy and quality of timbre while also minimizing recording artifacts such as proximity effect. The method has been implemented with renowned performers and orchestras, and has been used in the auralization of several new concert hall design projects.

This paper describes a method to record large musical ensembles for use in computer simulations and auralization. The method uses a multimicrophone hemispherical array distributed over an orchestra playing on a hard floor. A ray tracing program models the orchestra as sectors of a hemisphere corresponding to each microphone location. Microphone signals are then mapped to their respective source sectors by convolution with the model's outputs. The method permits recording and simulating an entire orchestra with its directivity, including the movement of the musicians while playing. The approach is geared towards achieving a high quality of orchestral musicianship and sense of orchestral ensemble while maintaining high spatial accuracy and quality of timbre while also minimizing recording artifacts such as proximity effect. The method has been implemented with renowned performers and orchestras, and has been used in the auralization of several new concert hall design projects.

Properly designed reference data and performance metrics can offer crucial aid to developers of advanced statistical recognition technologies. We focus here on audio data acquisition from close-talk, nearfield, and farfield sensors, and upon its processing, and its metrology. Our intention is to support the research community as it develops state of the art data acquisition and multimodal processing algorithms by supplying standard reference data, metrics, and sharable infrastructure, rather than developing the algorithms themselves.

The problem of source localization with ad hoc microphone networks in noisy and reverberant enclosures, given a training set of prerecorded measurements, is addressed in this paper. The training set is assumed to consist of a limited number of labelled measurements, attached with corresponding positions, and a larger amount of unlabelled measurements from unknown locations. However, microphone calibration is not required. We use a Bayesian inference approach for estimating a function that maps measurement-based feature vectors to the corresponding positions. The central issue is how to combine the information provided by the different microphones in a unified statistical framework. To address this challenge, we model this function using a Gaussian process with a covariance function that encapsulates both the connections between pairs of microphones and the relations among the samples in the training set. The parameters of the process are estimated by optimizing a maximum likelihood (ML) criterion. In addition, a recursive adaptation mechanism is derived where the new streaming measurements are used to update the model. Performance is demonstrated for 2-D localization of both simulated data and real-life recordings in a variety of reverberation and noise levels.

Localization of acoustic sources has attracted a considerable amount of research attention in recent years. A major obstacle to achieving high localization accuracy is the presence of reverberation, the influence of which obviously increases with the number of active speakers in the room. Human hearing is capable of localizing acoustic sources even in extreme conditions. In this study, we propose to combine a method based on human hearing mechanisms and a modified incremental distributed expectation-maximization algorithm (IDEM). Rather than using phase difference measurements that are modeled by a mixture of complex-valued Gaussians, as proposed in the original IDEM framework, we propose to use time difference of arrival (TDoA) measurements in multiple subbands and model them by a mixture of real-valued truncated Gaussians. Moreover, we propose to first filter the measurements in order to reduce the effect of the multi-path conditions. The proposed method is evaluated using both simulated data and real-life recordings.

The noise produced by complex systems such as automobiles, airplanes, machines, etc. contributes to overall noise of an environment. In the way to reduce these levels, the identification of the prominent sources is necessary. In many cases the beamforming technique is capable to carry out this task creating a visual map of the sound source. The goal of this work will be to give a basic understanding of beamforming techniques including software, hardware and to illustrate its wide applicability to common noise control and vibration problems. Resumo Os ruídos produzidos por sistemas complexos como automóveis, aeronaves, maquinas, etc. contribuem para o ruído global de um ambiente. De modo a reduzir estes níveis, a identificação das fontes proeminentes se faz necessária. Em muitos casos, a técnica de beamforming é capaz de realizar esta tarefa criando um mapa visual da fonte sonora. Este trabalho objetiva fornecer o entendimento básico sobre a técnica de beamforming, incluindo software, hardware e ilustrando sua ampla aplicação para o controle de ruído.

In this paper, we present a real-time convolutional neural network (CNN) based approach for speech source localization (SSL) using Android-based smartphone and its two built-in microphones under noisy conditions. We propose a new input feature set-using real and imaginary parts of the short-time Fourier transform (STFT) for CNN-based SSL. We use simulated noisy data from popular datasets that was augmented with few hours of real recordings collected on smartphones to train our CNN model. We compare the proposed method to recent CNN-based SSL methods that are trained on our dataset and show that our CNN-based SSL method offers higher accuracy on identical test datasets. Another unique aspect of this work is that we perform real-time inferencing of our CNN model on an Android smartphone with low latency (14 milliseconds(ms) for single frame-based estimation, 180 ms for multi frame-based estimation and frame length is 20 ms for both cases) and high accuracy (i.e. 88.83% at 0dB SNR). We show that our CNN model is rather robust to smartphone hardware mismatch, hence we may not need to retrain the entire model again for use with different smartphones. The proposed application provides a 'visual' indication of the direction of a talker on the screen of Android smartphones for improving the hearing of people with hearing disorders. INDEX TERMS Convolutional neural network, speech source localization (SSL), smartphone, direction of arrival (DOA).

In this paper, we propose a data-driven approach for the reconstruction of unknown room impulse responses (RIRs) based on the deep prior paradigm. We formulate RIR reconstruction as an inverse problem. More specifically, a convolutional neural network (CNN) is employed prior, in order to obtain a regularized solution to the RIR reconstruction problem for uniform linear arrays. This approach allows us to avoid assumptions on sound wave propagation, acoustic environment, or measuring setting made in state-of-the-art RIR reconstruction algorithms. Moreover, differently from classical deep learning solutions in the literature, the deep prior approach employs a per-element training. Therefore, the proposed method does not require training data sets, and it can be applied to RIRs independently from available data or environments. Results on simulated data demonstrate that the proposed technique is able to provide accurate results in a wide range of scenarios, including variable direction ...

A regression source location algorithm is developed to locate microcracks in bovine bone samples. Bone samples measuring 60 mm long, 22 mm wide and 5.5 mm thick were harvested from bovine femora. Artificial acoustic emission sources were created on the bone samples. The known location of these acoustic emission sources and the recorded arrival times at four acoustic emission sensors were used to develop regression equations, which when given new arrival time data could compute the acoustic emission source location. The regression equations were tested to determine if they could locate artificial acoustic emission sources and then acoustic emission emanating from microcracks in three-point bend tests. Results for the artificial acoustic emission source tests gave a mean absolute error of 0.69 mm and 0.71 mm for the X and Y coordinates respectively. In three-point bend tests a good spatial correlation was observed between the located microcracks and the observed fracture location.

In this paper, we examine three problems that rise in the modern, challenging area of far-field speech processing. The developed methods for each problem, namely (a) multichannel speech enhancement, (b) voice activity detection, and (c) speech recognition, are potentially applicable to a distant speech recognition system for voice-enabled smart home environments. The obtained results on real and simulated data, regarding the smart home speech applications, are quite promising due to the accomplished improvements made in the employed signal processing methods.

In preparation for upcoming wind tunnel acoustic experiments of a Hybrid Wing Body (HWB) vehicle with two jet engine simulator units, a series of twin jet aeroacoustic investigations were conducted leading to increased understanding and risk mitigation. A previously existing twin jet nozzle system and a fabricated HWB aft deck fuselage are combined for a 1.9% model scale study of jet nozzle spacing and jet cant angle effects, elevon deflection into the jet plume, and acoustic shielding by the fuselage body. Linear and phased array microphone measurements are made, and data processing includes the use of DAMAS (Deconvolution Approach for the Mapping of Acoustic Sources). Closely-spaced twin jets with a 5° inward cant angle exhibit reduced noise levels compared to their parallel flow counterparts at similar and larger nozzle spacings. A 40° elevon deflection into the twin jet plume, which is required for HWB ground rotation, can significantly increase upstream noise levels (more than 5 dB OASPL) with only minimal increases in the downstream direction. Lastly, DAMAS processing can successfully measure the noise source distribution of multiple shielded jet sources.

The problem of localizing and quantifying acoustic sources from a set of acoustic measurements has been addressed, in the last decades, by a huge number of scientists, from different communities (signal processing, mechanics, physics) and in various application fields (underwater, aero, or vibro acoustics). This led to the production of a substantial amount of literature on the subject, together with the development of many methods, specifically adapted and optimized for each configuration and application field, the variety and sophistication of proposed algorithms being sustained by the constant increase in computational and measurement capabilities. The counterpart of this prolific research is that it is quite tricky to get a clear global scheme of the state of the art. The aim of the present work is to make an attempt in this direction, by proposing a unified formalism for different well known imaging techniques, from identification methods (acoustic holography, equivalent sources, Bayesian focusing, Generalized inverse beamforming.. .) to beamforming deconvolution approaches (DAMAS, CLEAN). The hypothesis, advantages and pitfalls of each approach will be established from a theoretical point of view, with a particular effort in trying to separate differences in the problem definition (a priori information, main assumptions) and in the algorithms used to find the solution. Numerical simulations will be proposed for different source configurations (coherent/incoherent/extended/sparse distributions), and an experimental illustration on a supersonic jet will be finally discussed.

This paper presents some applications of advanced acoustic imaging techniques to flight test data. Engine jet is one of the main sources contributing to cabin noise in cruise flight conditions. The use of fuselage microphone arrays offers the possibility to extract information about the noise source mechanisms responsible for the jet noise, in terms of localization, quantification and directivity. Previous studies present beamforming space-frequency maps allowing the localization of main sources of noise along the jet axis as a function of the frequency. Beamforming is a robust acoustic imaging method, well adapted to the measurement conditions of fuselage microphones (low SNR), however it suffers from several drawbacks (space resolution, source correlations) that limits the analysis possibilities. In the present work, some advanced acoustic imaging techniques are implemented in order to overcome limitations inherent to beamforming. Deconvolution and identification methods are used to draw 1D and 2D source maps, offering the possibility to go deeper in the source characterization and to draw some physical interpretations about jet noise source mechanisms.

Phased microphone arrays have become a well-established tool for performing aeroacoustic measurements in wind tunnels (both open-jet and closed-section), flying aircraft, and engine test beds. This paper provides a review of the most wellknown and state-of-the-art acoustic imaging methods and recommendations on when to use them. Several exemplary results showing the performance of most methods in aeroacoustic applications are included. This manuscript provides a general introduction to aeroacoustic measurements for non-experienced microphone-array users as well as a broad overview for general aeroacoustic experts.

During the last decades, vibration analysis has been used to evaluate condition monitoring and fault diagnosis of complex mechanical systems. The problem associated with these analysis methods is that the employed sensors must be in contact with the vibrant surfaces. To avoid this problem, the current trend is the analysis of the noise, or the acoustic signals, which are directly related with the vibrations, to evaluate condition monitoring and/or fault diagnosis of mechanical systems. Both, acoustic and vibration signals, obtained from a system can reveal information related with its operation conditions. Using arrays formed by digital MEMS microphones, which employ acquisition/processing systems based on FPGA, allows creating systems with a high number of sensors paying a reduced cost. This work studies the feasibility of the use of acoustic images, obtained by an array with 64 MEMS microphones (8x8) in a hemianechoic chamber, to detect, characterize, and, eventually, identify fai...

This work addresses the solution of localizing and enhancing hands-free speech inside the car environment. Cars have different types of sounds from outside, co-passengers dialogue and noise. To provide better-quality speech, a microphone array-based beamforming technique is used. This research work proposes the method for selected source localization, source separation, and enhancement. An estimation of the direction of arrival (DOA) to localize the signal direction and preferred direction is selected for speech enhancement. The spiral and sine-cosine algorithm (SSCA) algorithm is combined with an adaptive least mean square to adapt the system for different environments. The algorithm is implemented in hardware and tested in a real-time car environment. The results showed significant improvement in signal-to-noise ratio (SNR) of 5.2 dB and perceptual evaluation of speech quality (PESQ) of 2.3. Finally, the model is fine-tuned for the car to get better quality. The proposed technique is efficient, and results are compared with existing methods.

Convolutional Neural Networks (CNNs) are a popular choice for estimating Direction of Arrival (DoA) without explicitly estimating delays between multiple microphones. The CNN method first optimises unknown filter weights (of a CNN) by using observations and ground-truth directional information. This trained CNN is then used to predict incident directions given test observations. Most existing methods train using spectrallyflat random signals and test using speech. In this paper, which focuses on single source DoA estimation, we find that training with speech or music signals produces a relative improvement in DoA accuracy for a variety of audio classes across 16 acoustic conditions and 9 DoAs, amounting to an average improvement of around 17% and 19% respectively when compared to training with spectrally flat random signals. This improvement is also observed in scenarios in which the speech and music signals are synthesised using, for example, a Generative Adversarial Network (GAN). When the acoustic environments during test and training are similar and reverberant, training a CNN with speech outperforms Generalized Cross Correlation (GCC) methods by about 125%. When the test conditions are different, a CNN performs comparably. This paper takes a step towards answering open questions in the literature regarding the nature of the signals used during training, as well as the amount of data required for estimating DoA using CNNs.