Search

Power transformers are the key equipment and the main sources of noise in smart substations. Noise is an important indicator that reflects the health and safety of transformers. Monitoring it is of great significance to the safe and stable operation of transformers. At the same time, clear noise data can help reduce public complaints about substation noise. This paper mainly analyses the causes of noise in smart substations from the core magnetostrictive phenomenon of transformers and obtains the basis for the design of noise monitoring devices in smart substations through the analysis of the causes. By analysing the principle of the noise monitoring terminal instrument, the A-weighting network for data correction is applied to obtain data that is relatively consistent with the human ear. To get higher accuracy, a method to overcome the overflow of data is proposed. Finally, sensors and analogue-to-digital converters are selected to design a noise monitoring terminal instrument. Thus, a noise monitoring system that is much more convenient than the existing monitoring system is established. The system was applied to a Smart Substation of 110kV level, and the sensitivity and accuracy have been verified through actual results.

Squeal noise is normally attributed to the self-excited vibration of the wheel induced by the lateral creepage when the wheel slides laterally on the top of rail. Due to its unique tone characteristics and high sound pressure level, the human ear is very sensitive to it. With the improvement of living standards, people's tolerance for squeal noise is getting lower and lower. In order to mitigate and control squeal noise, it is necessary to analyze the sensitivity of relevant parameters. In this paper, a prediction model is developed to analyze the effect of various parameters on the sound pressure level of squeal noise. The results show that squeal noise can be significantly reduced by decreasing angle of attack and rolling speed. In addition, the sound pressure of squeal noise can be reduced by reasonably controlling the negative slope of creepage. Furthermore, this mathematical model is used to perform a sensitivity analysis. The results show that squeal noise is most sensitive to the parameter of angle of attack, and the reason for this is illustrated. In particular, the analysis shows that modifying the slope of the creepage curve is effective in mitigating squeal noise, which provides a reference for mitigating and controlling squeal noise in practice, such as the application of friction modifiers at wheel/rail interface.

Tinnitus is an unusual sensation of hearing perceived without direct sound source. Given the multiplicity of management methods, a work of schematization of the most used therapies; i.e. Tinnitus Retraining Therapy (TRT), Cognitive and behavioral therapy (CBT), and Phase-out” sound therapy (POST) is led based on the Jastreboff model. Human ear models can be used to predict therapy effectiveness. Therefore, we presented three different models (an electric model, an acoustic model and a finite element model) which we then compared by applying them to two external ear shapes. The first is considered to have an auditory canal of uniform section and the second approaches the real shape of the external ear (variable section). The finite element (FE) model gives the most accurate results seeing that it allows to approach the real dimensions of the outer ear. The acoustic model gives results with a maximum deviation of 7,2 percent compared to the FE model with reduced computation time and simpler manner. The electric model is best suited if we consider the outer ear as a tube of uniform section closed at its end.

In the field of ear shape clustering, 3D ear modeling and ear related personal product implementation, it is of great significance to retrieve the accurate position of some key physiological curves and key points. Traditional edge extraction method is very sensitive to lighting and pose variations. A method for extracting the key physiological curves of the human ear based on an improved instance segmentation network YOLACT is proposed in this paper. By labeling the human ear dataset, the YOLACT model is trained. By improving the network prediction process, it can accurately segment different regions of the human ear and extract key physiological curves. Compared with the original YOLACT and other segmentation methods, it shows higher segmentation accuracy on the test dataset and can extract more accurate human ear physiological curves. The proposed method also shows more robustness under lighting variation, pose variation and slight occlusion.

In order to make full use of the noise information of substation power equipment and to deeply explore the value of noise, a voiceprint imaging technology in substation patrol was proposed in this paper which replace the the noise monitoring by human ear. The basic principle, algorithm, array selection and other technical details of the voiceprint imaging technology were introduced in detail. The effectiveness of the technical measure in substation patrol was proved by an application case. The voiceprint imaging technology in substation patrol is safe, fast and accurate. It fills the technical gap of power equipment noise patrol and greatly improves the efficiency of substation patrol.

The key to underwater target recognition is to extract the effective features of underwater target radiation noise. This paper presents an effective method for underwater target recognition and classification by extracting Mel-Frequency Cepstral Coefficients (MFCCs) features of underwater target radiation noise. Compared with traditional spectral analysis methods, MFCC makes full use of the non-linear auditory effect of the human ear with different perception capabilities for sounds of different frequencies. In this paper, the classification experiment of the radiated noise of the three types of measured underwater targets is done, where the MFCC feature vectors of the three types of targets are extracted, and the K-Nearest Neighbor (K-NN) algorithm is used to classify and identify them. Finally, the experimental results show that the method is effective.

A 24 GHz Doppler radar based stethoscope has been modified for non-contact diagnosis of heart sounds in a high-noise environment. By adding a digital single-sideband (SSB) modulator, the very low-frequency and difficult to hear heart sounds detected by this stethoscope were shifted toward the more sensitive part of the human ear’s frequency response.In a quiet environment, the heart sounds heard by the modified radio frequency (RF) stethoscope are shown to be the same as those heard by a conventional stethoscope. However, the modified RF stethoscope has an advantage in an extremely high noise environment, where a conventional stethoscope cannot be used. Under those conditions, the quality of the RF stethoscope’s acoustic response is shown to remain substantially unchanged.

Traditional on-line detection of partial discharge based on ultrasonic method has become an important means to detect the insulation condition of electrical equipment. However, most of its detection is based on the threshold value to realize the function similar to “traffic light”, that is, to detect whether partial discharge occurs or not. Gammatone frequency cepstral coefficient (GFCC) is based on human cochlear model and are widely used in voiceprint identification. This paper proposed an extraction algorithm of the characteristic parameter GFCC, and the PD ultrasonic signals of particle discharge and floating discharge are taken as examples, the GFCC of ultrasonic signals of those discharges are extracted. A preliminary partial discharge ultrasonic identification method based on GFCC is proposed. The method can realize on-line intelligent nonintrusive detection of the insulation state of high-voltage equipment, and break the limitation of traditional detection and diagnosis relying on certain threshold or human ear identification.

Audio Scene Classification (ASC) is a very important part of acoustic scene understanding. Recurrent Neural Network (RNN) models are widely adopted with proven successes in ASC. In this paper, the Gated Recurrent Unit (GRU) is introduced for ASC. The Mel-Frequency Cepstral Coefficients (MFCCs) which are closer to the auditory characteristics of the human ear are extracted as feature vectors and inputs to the classifier. The GRU is a slightly more simplified variation of the Long Short-Term Memory Unit (LSTM), and it is simpler than the standard LSTM. The Gated Recurrent Units (GRUs) are used to build a network model to implement ASC while controlling the number of parameters with respect to their performance. In these experiments, good results were achieved on the UrbanSound8Kdataset with an average accuracy of 83.8%, which has significant advantages over traditional network models.

The use of deep neural networks for speech recognition and recognizing speech commands continues to grow. This necessitates an understanding of the security risks that goes along with this technology. This paper analyzes the ability to interfere with the performance of neural networks for speech pattern recognition. With the methods proposed herein, it is a simple matter to create adversarial data by overlaying audio of a command at a fairly unnoticeable amplitude. This causes the neural network to lose around 20% accuracy and misidentify commands for other commands with an average to high confidence value. Such an attack is virtually undetectable to the human ear.

The accuracy of existing auditory saliency calculation models was analyzed in this paper. The subjective auditory attention saliency of 20 pieces of sound materials was judged by psychoacoustic experiment. Then the objective saliency of these sound materials was calculated using the existing auditory saliency calculation models. According to the relationship between the calculation results of the saliency calculation models and the subjective auditory attention perception of the human ear, the accuracy of existing auditory saliency calculation models was analyzed. The experimental results showed that, although auditory saliency computational models could calculate and judge the attention saliency of sound on certain extent, they had own shortcomings when calculating different types of sounds. Choosing different computational models based on the characteristics of the sound could improve the prediction accuracy of the auditory saliency calculation models.

Human ear recognition has attracted more attention in the past several years. In terms of the redundant information of Histogram of Oriented Gradient (HOG), this paper proposes a human ear recognition method in which the principal component analysis (PCA) was carried out on the HOG dimension reduction, then fused with local binary pattern (LBP) features. Using minimum distance classification for classification, experiments on USTB human ear image library shows that the proposed method improves the recognition rate and speeds up the training and detection.

Audio-based automated pulmonary symptom detection has the potential to offer accurate and continuous assessment of patients with lung disease. However, privacy preservation becomes a significant issue when it comes to continuous passive audio recording. Various techniques have been employed to obfuscate the speech within audio in these applications. However, that penalizes the accuracy of detection by affecting of-interest, non-speech audio parts. This is inevitably undesirable as it contradicts the notion of sensing. In this paper, we propose a novel algorithm to achieve the goal by employing a machine-learning-based vowel detection algorithm. The algorithm is implemented in a configurable manner to address many different scenarios of data collection. Logistic regression has been utilized to make the algorithm feasible for on-device implementation. We have shown that our algorithm achieves the goals in that the obfuscated speech is unrecognizable while cough sounds are identifiable by symptom detection models as well as a human ear.

Auricular branch of vagus nerve stimulation is continuously evolving as a minimal-invasive treatment option in patients suffering from chronic pain. For optimization of stimulation mechanisms, we conducted in-silico modelling of the electrode-tissue-nerve interface by a thorough literature review of the anatomical structures combined with nerve and vessel structures from episcopic images of a human ear. Four nerves and two vessels in the ear were implemented in a 3D CAD model for future optimization of stimulation regions and patterns. Mean diameters for nerves were 30 µm, and for vessels 100 µm. Mean distance between nerves and vessels was about 100 µm.

Human tissues have a high coefficient of absorption in the 60–300 GHz range. This focuses the study of human radiation exposure in this band to the skin and the cornea. The capacity of 60–300 GHz radiation to access the deeper parts of the ear has not been studied. Simulations show that over 90 % of the radiation that is presented parallel to the ear canal penetrates to 20mm into the canal at 60 and 90 GHz. The structures of the outer ear are highly protective, reducing penetration to 20%. At 300 GHz, the canal develops areas of up to 190% of the incident intensity, reducing to 150% when the outer ear structures are represented. These findings have important implications for the design of devices emitting 60–300 GHz radiation.

A Thai government's law prohibiting the cutting down of trees for making Thai xylophones (Ranā d xek) has had a major effect on musical instrument makers. It is necessary for these instrument makers to now synthesize or find a new replacement material. The original instrument must be used as a benchmark in terms of sound. An objective measurement procedure, beside the human ear, is needed to compare sounds of instruments made of the original wood and new materials. In this experimental project, audio measurements of the original Thai xylophone's sound were collected and statistically analyze, focusing on the inharmonic structure of its spectrum of partials. The result is shown as numerical data. This data can now be used as a benchmark and compared to possible replacement materials in further research.

Morphable models are useful shape priors for biometric recognition tasks. Here we present an iterative process of refinement for a 3D Morphable Model (3DMM) of the human ear that employs data augmentation. The process employs the following stages 1) landmark-based 3DMM fitting; 2) 3D template deformation to overcome noisy over-fitting; 3) 3D mesh editing, to improve the fit to manual 2D landmarks. These processes are wrapped in an iterative procedure that is able to bootstrap a weak, approximate model into a significantly better model. Evaluations using several performance metrics verify the improvement of our model using the proposed algorithm. We use this new 3DMM model-booting algorithm to generate a refined 3D morphable model of the human ear, and we make this new model and our augmented training dataset public.

In recent years, many researchers have shown interest in combining features of different biometrie traits to improve recognition performance of the biometrie systems. In this paper, we examine the feature-level fusion of two contactless biometric modalities of the same image i.e. ear and profile face. Initially, two most efficient local feature descriptors such as LDP (Local Directional Patterns) and LPQ (Local Phase Quantisation) are used to represent both biometric modalities. Due to combination of two feature descriptors, dimension of the feature sets are increased and so PCA is separately applied to both modalities before normalization and fusion steps. Finally, to obtain more discriminant nonlinear features the Kernel Dis-criminative Common Vector (KDCV) method is employed after fusion to the combined feature vector. Experimental evaluation on University of Notre Dame (Collection E) side face database clearly reveals the proposed method is more efficient to increase the recognition performance over other existing ear based unimodal and multimodal biometric systems.

This Paper is about finding a low cost solution to Audio effects used by musicians and audio engineers. From chorus to flanger, rotary, delay, overdrive, and pitch shifting, etc… This paper focuses on two effects: Delay and Pitch shifting. The first part the delay was used since the beginning of using electronics in musical instruments. It adds some delayed versions of the input signal which gives a warm tone and adds a depth to the sound of the instrument, from a psychological point of view it keeps hitting the ear drum of the listener to a level where the ear is used to it and actually is some strange type of a very short time addiction. It was done through using a circular buffer technique to save the memory and to get rid of latency. The second part is the pitch shifting. It actually allows one person to play two lines at the same time, giving the listener a variation of frequencies that hit the low pass filter which is the human ear, making some sort of linear surrounding effect depending on how the panning of the original signal and the shifted version is done. The techniques used are carefully selected and implemented with some types of interpolation to make sure to get the best results.

On the basis of the dereverberation method based on TF-GSC and minimum phase decomposition, considering the insensitivity of the human ear to the speech phase information, an improved method that only use the phase of a reverberant speech to replace the phase information of the whole component is proposed. The time domain waveforms produced by all the dereverberation methods have been comprehensively compared. At the same time, all the methods are evaluated by simulation using two subjective and three objective dereverberation evaluation indices. Results show that the improved method has better performance of dereverberation and save about half of the computational complexity so that the real-time processing speed of the algorithm is greatly improved.

This paper proposes a simple automatic 3D ear reconstruction method using a video selfie while having no knowledge about the scene. First we use the EXIF data stored within the images to estimate the intrinsic matrix. Then, standard structure from motion techniques are applied to obtain a reconstruction for every couple of images which will be merged into one reconstruction that represents the whole scene using an original procedure. Finally, the quasi-dense model is obtained by using the ASIFT and ZNCC descriptors. Our approach is simple to implement and only requires the use of a smart-phone camera. Also, the experimental results showed that this approach is very promising compared to the other methods used to solve this problem.

Ear has robust as well as reliable features, thus technique have been proposed for ear detection and recognition from 2D & 3D ear images. In the proposed work, the ring projection technique is used to convert a 2-D ear image into 1-D information. This ring projection technique requires less memory for storage, and is rotation invariant. The 1-D information, after normalization, is used for the construction of the semi-orthogonal wavelet. Using the corresponding semi-orthogonal wavelet of each ear image, 2-D ear image is decomposed to two levels; thus energy compaction ratio for each ear image is obtained. The proposed approach is tested on USTB ear image database. The database contains three images of each subject are captured (total 180 images), under different lighting conditions. They all are right ear images of the subjects. Then variations of ECR value with the various obstructions (ear-ring or partial hair occlusion) in ear illustrate the amount of useful information contained by each image after decomposition using wavelet.

Data mining has a great potential in different areas of health informatics. Data mining in health industry can minimize the health cost as well as reduces the risk of life by informing a person at initial stage. An automatic classification system capable of mining pathological data may contribute in health informatics significantly. In this paper, an automatic system to differentiate between pathological and normal data is developed. The developed system mines the pathological data on the basis of an acoustic analysis. The purpose of the acoustic analysis is to estimate the auditory spectrum of a voice sample by using the principle of the human auditory system called as critical bandwidths. The estimated auditory spectrum simulates the behavior of a human ear and acts like an expert clinician who can identify a pathological voice by hearing. The pathological data used for this study is recorded from the people who are suffering from more than 100 different types of voice disorders. Voice of a disordered patient feels noisy, harsh, strain, breathy and unpleasant to ears. During the training phase of the proposed system, it takes labeled normal and pathological data to generate acoustic models by using the Gaussian mixture model. While in deployment phase, an unknown and unlabeled voice sample is given to the system to determine its type, i.e. normal or pathological. The best obtained accuracy of the system is 99.50%.

This document belongs to the implementation of a system of recognition of the bases on the Hidden Models of Markov HMM. Due to a system of recognition, we will use the techniques of parameterization which handles the mechanisms of the human ear Mel Frequency cepstral coefficient MFCC and Perceptual linear prediction PLP starting from the database TIMIT. We also used two indices Jitter and Shimmer since they show very precise information about the voice of the person.

Morphological shape of the human ear presents a rich and stable information embedded on the curved 3D surface, which has invited lot attention from the forensic and engineer scientists in order to differentiate and recognize people. However, recognizing identity from morphological shape of the human ear in unconstrained environments, with insufficient and incomplete training data, dealing with strong person-specificity, and high within-range variance, can be very challenging. In this work, we implement a simple yet effective approach which uses and exploits recent local texture-based descriptors to achieve faster and more accurate results. Support Vector Machine (SVM) is used as a classifier. We experiment with two publicly available databases, which are IIT Delhi-1 and IIT Delhi-2, consisting of several ear benchmarks of different natures under varying conditions and imaging qualities. The experiments show excellent results beyond the state-of-the-art.

Healthcare experts have recently turned towards the use of Biosimulation models to understand the multiple or different causative factors that cause impairment in human organs. The applications of biosimulations have been applied in different biological systems ranging from human ear, cardiovascular, to neurovascular research using Finite Element Method (FEM). FEM provide a mathematical framework to simulate these dynamic biological systems. Visualizing and analyzing huge amounts of Finite Element (FE) Biosimulation numerical data is a strenuous task. In this paper, we demonstrate a Linked Data Visualiser -- called SIFEM Visualiser -- to help domain-experts to Visualise, analyze and compare biosimulation results from heterogeneous, complex, and high volume numerical data. The SIFEM Visualiser aims to help healthcare experts in exploring and analyzing Finite Element (FE) numerical data and simulation results obtained from different aspects of inner-ear (Cochlear) model - such as biological, geometrical, mathematical, and physical models. This publication has emanated from research supported in part by the research grant from Science Foundation Ireland (SFI) under Grant Number SFI/12/RC/2289 and EU project SIFEM (contract Number 600933). REFERE non-peer-reviewed

Infants' recognition has always been a huge challenge even in today's technology-driven world. Swapping, missing, kidnapping and mixing infants pose an exaggeration to the existing problem. While the advent of biometric tests has solved the problem of adult recognition, the similar thing for infants, could not serve an appropriate solution. The objective of this paper is to emphasize on the possibility of using human ear as an infant recognizing solution. The reason being this part of body attracted several types of research and studies to prove individuality and uniqueness. The major contributions of this paper includes (i) Implementation of seven algorithms on ear database of adult and infant, (ii) Comparison of the result of an algorithm, (iii) Preparation of infant database.

Biometrics have lately been receiving attention in popular media. Biometrics deal with identification and verification of individuals based on their behavioral or physiological characteristics. Biometrics will become one of the most important ways of the identification technology. Ear recognition might be a good solution since ear is visible, ear images are easy to be taken, and the ear structure does not change radically over time. In this paper an algorithm based on SIFT features for ear recognition is proposed. SIFT key points are extracted from ear image and an augmented vector of extracted SIFT features are created for matching. Firstly, a pre-processing phase is done by converting image to gray level. Then a median filter is applied to smooth the image and to remove noise if found. Edge detection is used for cropping ear part from the image. Then the SIFT features were extracted from ear image. Finally, the extracted features were classified by using minimum distance classifier. This method is invariant to scaling, translation and rotation. The experimental results showed that the proposed approach gives better results compared with other researchers and obtained over all accuracy almost 95.2 % for IIT Delphi database and 100% for AMI database.

External shape of the human ear presents a rich and stable information embedded on the curved 3D surface, which has invited lot attention from the forensic and engineer scientists in order to differentiate and recognize people. However, recognizing identity from external shape of the human ear in unconstrained environments, with insufficient and incomplete training data, dealing with strong person-specificity, and high within-range variance, can be very challenging. In this work, we implement a simple yet effective approach which uses and exploits recent local texture-based descriptors to achieve faster and more accurate results. Support Vector Machines (SVM) are used as a classifier. We experiment with two publicly available databases, which are IIT Delhi-1 and IIT Delhi-2, consisting of several ear benchmarks of different natures under varying conditions and imaging qualities. The experiments show excellent results beyond the state-of-the-art.

Biometrics includes the study of automatic methods for distinguishing human beings based on physical or behavioral traits. Finding good biometric methods has been researched extensively in recent years. Among several biometric features, ear is quite stable because it does not vary with age and emotion. The ear recognition works based on the height of the ear, reference line cut point and corresponding angles. The study is performed on the ear in random orientation and shows a greater accuracy than existing dominant approach. The recognition accuracy is increased by removing the noise in captured ear images and developing new methodologies to work with online images.

The purpose of this study is to develop a system that enables location finding of a small sound. The system is composed of a biologically-inspired system which uses a hearing mechanism based on the human ear and a mechanism for perceiving weak signals that uses stochastic resonance. Using this proposed system, we implemented the location finding of small sounds through numerical simulations and hardware experiments. Good results were obtained for the small sound source location finding.

Julian Haegele,1 Robert L Duschka,1 Matthias Graeser,2 Catharina Schaecke,1 Nikolaos Panagiotopoulos,1 Kerstin Lüdtke-Buzug,2 Thorsten M Buzug,2 Jörg Barkhausen,1 Florian M Vogt1 1Clinic for Radiology and Nuclear Medicine, 2Institute of Medical Engineering, University Hospital Schleswig Holstein, Lübeck, Schleswig-Holstein, Germany Background: Magnetic particle imaging (MPI) uses magnetic fields to visualize superparamagnetic iron oxide nanoparticles (SPIO). Today, Resovist® is still the reference SPIO for MPI. The objective of this study was to evaluate the in vivo blood half-life of two different types of Resovist (one from Bayer Pharma AG, and one from I’rom Pharmaceutical Co Ltd) in MPI. Methods: A Resovist concentration of 50µmol/kg was injected into the ear artery of ten New Zealand White rabbits. Five animals received Resovist distributed by I’rom Pharmaceutical Co Ltd and five received Resovist by Bayer Pharma AG. Blood samples were drawn before and directly after injection of Resovist, at 5,10, and 15minutes, and then every 15­minutes until 120minutes after the injection. The MPI signal of the blood samples was evaluated using magnetic particle spectroscopy. Results: The average decline of the blood MPI signal from the two distributions differed significantly (P=0.0056). Resovist distributed by Bayer Pharma AG showed a slower decline of the MPI signal (39.7% after 5minutes, 20.5% after 10minutes, and 12.1% after 15minutes) compared with Resovist produced by I’rom Pharmaceutical Co Ltd (20.4% after 5minutes, 7.8% after 10minutes, no signal above noise level after 15minutes). Conclusion: In MPI, the blood half-life of an SPIO tracer cannot be equalized to the blood half-life of its MPI signal. Resovist shows a very rapid decline of blood MPI signal and is thus not suitable as a long circulating tracer. For cardiovascular applications in MPI, it may be used as a bolus tracer. Keywords: magnetic particle imaging, superparamagnetic iron oxide nanoparticles, magnetic particle spectroscopy, blood half-life time, Resovist&reg

Biometrics includes the study of automatic methods for distinguishing human beings based on physical or behavioural traits. Finding good biometric features and recognition methods has been researched extensively in recent years. Among several biometric features ear is quite stable because it does not vary with age and emotion. However, very few researches have been accomplished with this biometric. Hence, this paper proposes a new approach to recognize 2d human ear based on height of ear, reference line cut point and corresponding angles. Experimental study shows that this paper has a greater accuracy then existing dominant approach.

Echo hiding is a digital audio watermarking technology which uses the human ear's time-domain masking effects. In this paper, we proposed a new echo kernel with several echoes as an alternative to the single echo in conventional echo hiding. The proposed method introduced multiple small echoes instead of a single echo, the total energy of echo pulses will be distributed to each small component of the echo. For every single echo, the magnitude will be reduced, and it carries less energy. But the total energy embedded in increases, with the premise of ensuring the imperceptibility of the original audio. Therefore, the proposed method is more robust to common signal processing operations and resistant to tampering, comparing with the case of conventional echo hiding method. Also, in subjective evaluation the proposed method showed quite good perceptual quality.

In this paper, the effects of electromagnetic (EM) absorption on human ear and head shapes has been investigated. Different types of human head models such as spherical, cubical, and realistic human head models were used in this paper. Also, different types of ear shapes were considered with the cubical, spherical, and head models whereas a low-loss thin ear, or a lossy realistically shaped ear, was used with the realistic head models. Commercially available software CST Microwave Studio based on the finite-difference time-domain (FDTD) method was used for the numerical simulations. The SAR circulation depends appreciably on the size and shape of the ear, despite of the head model. The head shapes are fewer affected than the ear shape. Comparison studies of SAR bring in the realistic human head for homogeneous and inhomogeneous cases are also investigated. Results, it can be seen that the local maximum SAR motivated in the homogeneous human head model is larger than that induced in the inhomogeneous human head model. Also it is demonstrated that, the maximum local SARs in the head tissue with the low-loss thin ear are larger than those with the lossy realistically shaped ear.

This paper introduces a novel set of non-linear spectro-temporal features that improve automatic speech recognition performance in presence of room reverberation. The solution is based on extracting features derived from auditory characteristics, which include gammatone filtering, non-linear processing and modulation spectral processing to emulate the mechanisms performed in the cochlea and middle ear aimed to improve robustness in human ear. Experiments are performed on Aurora-5 meeting recorder digit task (mrd), captured with four different distant microphones in hands-free mode at a real meeting room. For comparison purposes the recognition results obtained using standard conventional features are tested. The experimental results show that the proposed features provide considerable improvements with respect to state of the art feature extraction techniques.

Aimed at the influence of illumination and posture changed to the human ear recognition, isometric mapping (Isomap) algorithm of manifold is applied to the feature extraction stage of human ear recognition. The results of experiment indicate that using Isomap algorithm in the human ear recognition, can reduce the influence on rate when the posture and illumination changed, and improve the recognition rate.

The Human Ear is a new class of relatively stable Biometrics that has drawn researcher's attention recently. Human ear is a perfect data for passive person identification, which can be applied to provide security in the public places. In this article, we overview the proposed Pattern extraction algorithms from 2D and 3D Ear images.

This paper presents the observation from SAR images of the dried Lop Nur Lake in Xinjiang of west China. Lop Nur Lake is one of the driest places in the world and finally lost its last drop of water in 1972. It is well known for its "human ear shape" feature in optical remote sensing images. Likewise, the "ear" feature is shown in synthetic aperture radar (SAR) images, and is even larger because of the penetration effect. The universal existence of lacustrine deposit with high water content and salinity are undetectable by previous remote sensing images in Lop Nur Lake. We have made some laboratory experiments about subsurface soils and developed an improved dielectric model for moist saline soil at microwave bands. A field trip was conducted from October 25 to November 8, 2006 to validate the image interpretation results. As an important indicator for environment degradation and climate change study in arid region, Lop Nur does need long-term and overall research to answer many scientific questions.

Due to its semi-rigid shape and robustness against change over time, the ear has become an increasingly popular biometric feature. It has been shown that combining individual biometric methods into multi-biometric systems improves recognition. What features should be used, how they should be captured, what algorithms should be used, and how they should be combined are all open questions. In this paper, we discuss several existing methods of combination and the recognition rates of each.

We introduce an unobtrusive sensor-based control system for human-machine interface to control robotic and rehabilitative devices. The interface is capable of directing assistive robotic devices in response to tongue movement and/or speech without insertion of any device in the vicinity of the oral cavity. The interface is centered on the unique properties of the human ear as an acoustic output device. Our work has shown that various movements within the oral cavity create unique, traceable pressure changes in the human ear, which can be measured with a simple sensor (such as a microphone) and analysed to produce commands signals, which can in turn be used to control robotic devices. In this work, we present: 1) an analysis of the sensitivity of human ear canals as acoustic output device, 2) the design of a new sensor for monitoring airflow in the aural canal, 3) pattern recognition procedures for recognition of both speech and tongue movement by monitoring aural flow across several human test subjects, and 4) a conceptual design and simulation of the machine interface system

Because of the need for automatic, easily repeated physiological measurements in animal models of chronic cardiac disease states, we have developed a rapid and accurate, optically assisted method for determining blood pressure in New Zealand White rabbits. The method relies on the change in the attenuation of laser light through a peripheral ear artery as a function of an externally applied pressure ramp.

Two technologies, 3-D reconstruction of human temporal bones and finite element analysis of ear biomechanics, were combined to construct a computational model for describing function of the ear for sound transmission. A set of histological sections of a human temporal bone was prepared, scanned, and digitized for imaging analysis. The aligned and treated images of the middle ear components, external ear canal and middle ear cavity were constructed with attached ligaments, muscles, and tendons. A 3-D CAD model was first created which provided accurate geometric data. Based on the solid model, the finite element (FE) model of the human ear was generated with whole middle ear and ear canal. The dynamic behavior of the FE model was very close to the actual mechanical performance of human temporal bone with 90 dB SPL at the eardrum across auditory frequencies. The model will be further used for acoustic analysis of the ear. (Supported by Whitaker Foundation and OCAST).

A system is described for the analysis of recordings made on infants suffering from a condition of the airway known as stridor. The method is based on the use of a digital filterbank designed to model the critical band response of the human ear. This physiologically based filterbank uses 145 fourth-order bandpass filters whose center frequencies are spaced on a perceptual scale as determined by the critical band theory over the range 100 Hz to 6.4 kHz. Results from three recordings are illustrated and discussed. >

Transform or subband audio coders can deliver high quality reconstruction at rates around two bits per sample. Most quantization strategies take into account masking properties of the human ear to make the quantization noise less noticeable. In this paper we describe a new coder in which we extend such quantization strategies by incorporating run-length and arithmetic encoders. They lead to improved performance for quasi-periodic signals, including speech. The quantization tables are computed from only a few parameters, allowing for a high degree of adaptability without increasing quantization table storage. To improve the performance for transient signals, the coder uses a nonuniform modulated lapped biorthogonal transform with variable resolution without input window switching. Experimental results show that the coder can be used for good quality signal reproduction at rates close to one bit per sample, and quasi-transparent reproduction at two bits per sample.

Traditional spectrography commonly found in speech sciences laboratories, makes use of fixed bandwidth bandpass analysis filters. The bandwidth can be adjusted to optimise between time and frequency resolution of the output spectrogram as desired. Acoustic analysis by the human ear can be modelled as a bank of bandpass filters whose bandwidth varies as a function of centre frequency. We describe a real-time transputer-based implementation of a spectrograph based on knowledge of peripheral human hearing. The purpose of this paper is to compare and contrast the output from this system with traditional spectrograms for a variety of musical performances to illustrate: (i) the differences between the systems, (ii) psychoacoustic ideas relating to music, and (iii) the complex nature of the signal that is sent to higher centres of neural processing from each ear.

We propose an algorithm to modify the synchronized overlap-and-add (SOLA) technique. SOLA is a popular technique for time scale modification of speech and audio signal. It changes the time scale of the signal while maintaining the pitch information. However, when the time scale of the signal is compressed more, the synthesized speech of SOLA has high articulations rate that it becomes almost impossible for the human ear to comprehend. The proposed algorithm can enhance the quality if the time scaled signal by using a time varying time scale factor which is locally adapted to the input signal. It is found that the intelligibility of the synthesized signal using the proposed algorithm is considerably better than the one synthesized from simple SOLA.