Study on Acoustic Modeling in a Mandarin Continuous Speech Recognition

The design of acoustic models is of vital importance to build a reliable connection between acoustic wave-form and linguistic messages in terms of individual speech units. According to the characteristic of Chinese phonemes, the base acoustic phoneme units set is decided and refined and a decision tree based state tying approach is explored. Since one of the advantages of top-down tying method is flexibility in maintaining a balance between model accuracy and complexity, relevant adjustments are conducted, such as the stopping criterion of decision tree node splitting, during which optimal thresholds are captured. Better results are achieved in improving acoustic modeling accuracy as well as minimizing the scale of the model to a trainable extent.

Parallel optimization of underwater acoustic models:A survey

Underwater acoustic models are effective tools for simulating underwater sound propagation.More than 50 years of research have been conducted on the theory and computational models of sound propagation in the ocean.Unfortunately,underwater sound propagation models were unable to solve practical large-scale three-dimensional problems for many years due to limited computing power and hardware conditions.Since the mid-1980s,research on high performance computing for acoustic propagation models in the field of underwater acoustics has flourished with the emergence of high-performance computing platforms,enabling underwater acoustic propagation models to solve many practical application problems that could not be solved before.In this paper,the contributions of research on high-performance computing for underwater acoustic propagation models since the 1980s are thoroughly reviewed and the possible development directions for the future are outlined.

Microstructure Features and the Macroscopic Acoustic Behavior of Gassy Silt in the Yellow River Delta

The morphological changes in isolated bubbles in gassy silt play a critical role in the microscopic structures between soil particles and bubbles and macroscopic physical properties.Based on X-ray CT scanning experiments under various vertical loads(four levels),self-designed acoustic macro experiments,and a series of formula revisions to the macro-air-bearing silt sound-velocity prediction model,this paper discusses the macro-and micro-scale features of gassy silts from the Yellow River Delta.The samples consisted of different proportions of silt from the Yellow River Delta and porous media,and they were used to form two types of aerosol silts with initial gas contents of 4.23%and 7.67%.The results show that the air bubble content and external load considerably affect the microstructural parameters and acoustic behavior of gassy silt in the Yellow River Delta.The macroscopic sound velocity showed a linear positive correlation with vertical load and relation to microstructural parameters in varying manners and degrees.Based on the traditional Biot-Stoll acoustic model,the gas-phase medium coefficient was introduced for the proper calculation and prediction of the sound velocity of air-bearing silt.The errors of the overall prediction varied between 5.6%and 9.6%.

DNN-HMM based acoustic model for continuous pig cough sound recognition

To detect the respiratory disease through pig cough sound in the early stage,a novel method based on Deep Neural Networks-Hidden Markov Model(DNN-HMM)was proposed to construct an acoustic model for continuous pig cough sound recognition.Noises in the continuous pig sounds were eliminated by the Wiener algorithm based on wavelet thresholding the multitaper spectrum,and the experimental corpus was obtained from the denoised continuous pig sounds.The 39-dimensional Mel Frequency Cepstral Coefficients(MFCC)extracted from the corpus were considered as feature vectors.Sounds in pig farms were divided into pig coughs,non-pig coughs,and silence segments.In the HMM,the number of hidden states of pig cough,non-pig cough and silence segments were 5,5 and 3 respectively,and the observation states represented the feature vectors of the continuous pig sound signal.Based on experiments and empirical theory,the DNN model with 3 hidden layers and 100 nodes per layer was used to describe the correspondence between hidden states and observation serials.Through experiments,the context frames of DNN input were set to 5.Under the condition of optimal parameter setting,the traditional acoustic model Gaussian Mixture Model-Hidden Markov Model(GMM-HMM)was compared with DNN-HMM through a 5-fold cross-validation experiment.It was found that the Word Error Rate(WER)of each group in DNN-HMM was lower than that in GMM-HMM,and the average WER was 3.45%lower.At the same time,the best result of the DNN-HMM model was obtained with the lowest WER of 7.54%,and the average WER was 8.03%.The results showed that the method of DNN-HMM based acoustic model for continuous pig cough sound recognition was stable and reliable.

AN ACOUSTIC MODELING STUDY OF FRICATION LEVELS OF HIGH VOWELS IN CHINESE

Frication of high vowels has three levels in auditory impression,including non-frication,low frication,and high frication.The aim of this study was to study the acoustic characteristics and social variability of the three levels via the method of multivariate statistical modeling.We estimated the weighting of each parameter by fitting the model of Boosted Logistic Regression to obtain the optimal parametric set.Pattern recognition then quantified the social variability with the optimal set.It showed that the acoustic characteristics of the vowel frication are mainly reflected in the formants,of which the second bandwidth (B2) plays the most prominent role in distinguishing the three levels.Three dimensional diagrams further illustrated that the high frication group has the greatest B2 value,with the smallest first formant (F1) and skewness,while the non-frication group is completely the opposite.Results of pattern recognition showed that the degrees of variation in age and gender are both close to 19%,where the higher the frication level is,the greater the variability is.This paper thus suggested that low frication variants should be treated as fricative approximants,instead of fricative vowels as reported in the literature.

Experiment, acoustic model for the self-oscillation of coaxial swirl injector and its influence to combustion of liquid rocket engine

During the experiment of gas/liquid coaxial swirl injector conducted with air and water under atmosphere environment, it is observed that the injector may selfoscillate. The self oscillation periodically occurs and vanishes with the increasing velocity of the gas flow.A theoretical model is presented based on the experiment investigation. Simulation of the acoustic process has been performed and conclusions consistent with the experiment can be drawn from the theoretical model, which explains the exPeriment phenomena quite well. At last, the comparison between phenomena of the self oscillation and some experiments of LRE indicates that some instability phenomena in oxygen/hydrogen propellant rocket engine may be the related to self oscillation in coaxial injectors

Geoacoustic model and acoustic reflection properties of fluid mud layer in Changjiang Estuary and Hangzhou Bay

A generalized geoacoustic model of fluid mud layer in Chanaiiang Estuary and Hangzhou Bay has been derived from a large amount of in-situ measurements of bulk density (p) profiles of the lay6rs and of lab measurements of acoustic velocities (c) and attenuation coefficients (o) of the fluid mud samples with different values of p for four frequencies of 100 kHz, 150 kHz, 500 kHz, 1500 kHz. The main features of the geoacoustic model can be expressed as follows: from the upper boundary, the bulk density of the fiuid mud increases linearly with depth z, however there is a gradient change (knee) when p is about 12.5 kN/m', then p increases linearly to a value about 15.0 kN/m'. After p more than 15.0, the fluid mud layer quickly transform into an ooze layer. In the fluid mud layer, the acoustic velocity c can be regarded as constant since its variation with z less than 1.5%, and a minimum vaue of c ekists when p is about 13.5 kN/m'. The variations of β with p and with frequency f are linear. Based on the geo-acoustic model and the ray theory, simulations of sound refiection from the fluid mud layers have been made, and some significallt results obtained, from which the bulk density profiles of fluld mud layers can be derived inversely.

Numerical analysis of the resonance mechanism of the lumped parameter system model for acoustic mine detection

The method of numerical analysis is employed to study the resonance mechanism of the lumped parameter system model for acoustic mine detection. Based on the basic principle of the acoustic resonance technique for mine detection and the characteristics of low-frequency acoustics, the “soil-mine” system could be equivalent to a damping “mass-spring” resonance model with a lumped parameter analysis method. The dynamic simulation software, Adams, is adopted to analyze the lumped parameter system model numerically. The simulated resonance frequency and anti-resonance frequency are 151 Hz and 512 Hz respectively, basically in agreement with the published resonance frequency of 155 Hz and anti-resonance frequency of 513 Hz, which were measured in the experiment. Therefore, the technique of numerical simulation is validated to have the potential for analyzing the acoustic mine detection model quantitatively. The influences of the soil and mine parameters on the resonance characteristics of the soil–mine system could be investigated by changing the parameter setup in a flexible manner.

A Modified Monte Carlo Model of Speckle Tracking of Shear Wave Induced by Acoustic Radiation Force for Acousto-Optic Elasticity Imaging

A modified Monte Carlo model of speckle tracking of shear wave propagation in scattering media is proposed. The established Monte Carlo model mainly concerns the variations of optical electric field and speckle. The two- dimensional intensity distribution and the time evolution of speckles in different probe locations are obtained. The fluctuation of speckle intensity tracks the acoustic-radiation-force shear wave propagation, and especially the reduction of speckle intensity implies attenuation of shear wave. Then, the shear wave velocity is estimated quantitatively on the basis of the time-to-peak algorithm and linear regression processing. The results reveal that a smaller sampling interval yields higher estimation precision and the shear wave velocity is estimated more efficiently by using speckle intensity difference than by using speckle contrast difference according to the estimation error. Hence, the shear wave velocity is estimated to be 2.25 m/s with relatively high accuracy for the estimation error reaches the minimum （0.071）.

A Theoretical Model for the Asymmetric Transmission of Powerful Acoustic Wave in Double-Layer Liquids

A theoretical model which couples the oscillation of cavitation bubbles with the equation of an acoustic wave is utilized to describe the sound fields in double-layer liquids, which can be used to realize the asymmetric transmission of acoustic waves. Numerical simulations show that the asymmetry is related to the properties of the host liquids and the input acoustic wave. Asymmetry can be enhanced if the maximum number density or the ambient radius of the cavitation bubbles in the low cavitation threshold liquid increases. Moreover, the direction of rectification will be reversed if the amplitude of the input acoustic wave becomes high enough.

Geodesic acoustic mode in a reduced two-fluid model

A reduced two-fluid model is constructed to investigate the geodesic acoustic mode（GAM）. The ion dynamics is sufficiently considered by including an anisotropic pressure tensor and inhibited heat flux vector, whose evolutions are determined by equations derived from the 16-momentum model. Electrons are supposed to obey the Boltzmann distribution responding to the electrostatic oscillation with near ion acoustic velocity. In the large safety factor limit, the GAM frequency is identical with the kinetic one to the order of 1 q2 when zeroing the anisotropy. For general anisotropy, the reduced two-fluid model generates the frequency agreeing well with the kinetic result with arbitrary electron temperature. The present simplified fluid model will be of great use and interest for young researchers and students devoted to plasma physics.

Phoneme Sequence Modeling in the Context of Speech Signal Recognition in Language “Baoule”

This paper presents the recognition of “Baoule” spoken sentences, a language of C?te d’Ivoire. Several formalisms allow the modelling of an automatic speech recognition system. The one we used to realize our system is based on Hidden Markov Models (HMM) discreet. Our goal in this article is to present a system for the recognition of the Baoule word. We present three classical problems and develop different algorithms able to resolve them. We then execute these algorithms with concrete examples.

An equivalent-tool theory for acoustic logging and applications

The influence of an acoustic logging tool on borehole guided wave propagation should be considered in the processing and inversion of the guided waves for formation acoustic property estimation. This study introduces an equivalent-tool theory that models the tool response using an elastic rod with an effective modulus and applies the theory to multipole acoustic logging for both wireline and logging while drilling （LWD） conditions. The theory can be derived by matching the tool’s acoustic impedance/conductance to that of the multipole acoustic wavefield around the tool, assuming that tool radius is small compared to wavelength. We have validated the effectiveness and accuracy of the theory using numerical modeling and its practicality using field data. In field data applications, one can calibrate the tool parameters by fitting the theoretical dispersion curve to field data without having to consider the actual tool’s structure and composition. We use a dispersion correction example to demonstrate an application of the simple theory to field data processing and the validity of the processing result.

Non-invasive determination of hepatic steatosis by acoustic structure quantification from ultrasound echo amplitude

AIM:To use leptin-deficient(ob/ob) mice with demonstrated differences in steatosis levels to test a new diagnostic method using the acoustical structure quantification(ASQ) mode and the associated analytical parameter,"focal disturbance ratio"(FD-ratio).METHODS:Nine ob/ob mice,at 5,8,and 12 wk of age(n = 3 in each age group),were used as models for hepatic steatosis.Echo signals obtained from ultrasonography in the mice were analyzed by ASQ,which uses a statistical analysis of echo amplitude to estimate inhomogeneity in the diagnostic region.FD-ratio,as calculated from this analysis,was the focus of the present study.FD-ratio and fat droplet areas and sizes were compared between age groups.RESULTS:No fibrosis or inflammation was observed in any of the groups.The fat droplet area significantly(P < 0.01) increased with age from 1.25% ± 0.28% at 5 wk to 31.07% ± 0.48% at 8 wk to 51.69% ± 3.19% at 12 wk.The median fat droplet size also significantly(P < 0.01) increased with age,from 1.33(0.55-10.52) m at 5 wk,2.82(0.61-44.13) m at 8 wk and 6.34(0.66-81.83) m at 12 wk.The mean FD-ratio was 0.42 ± 0.11 at 5 wk,0.11 ± 0.05 at 8 wk,and 0.03 ± 0.02 at 12 wk.The FD-ratio was significantly lower at 12 wk than at 5 wk and 8 wk(P < 0.01).A significant negative correlation was observed between the FD-ratio and either the fat droplet area(r =-0.7211,P = 0.0017) or fat droplet size(r =-0.9811,P = 0.0052).CONCLUSION:This tool for statistical analysis of signals from ultrasonography using the FD-ratio can be used to accurately quantify fat in vivo in an animal model of hepatic steatosis,and may serve as a quantitative biomarker of hepatic steatosis.

Design of multi-layered porous fibrous metals for optimal sound absorption in the low frequency range

We present a design method for calculating and optimizing sound absorption coefficient of multi-layered porous fibrous metals （PFM） in the low frequency range. PFM is simplified as an equivalent idealized sheet with all metallic fibers aligned in one direction and distributed in periodic hexagonal patterns. We use a phenomenological model in the literature to investigate the effects of pore geometrical parameters （fiber diameter and gap） on sound absorption performance. The sound absorption coefficient of multi- layered PFMs is calculated using impedance translation theorem, To demonstrate the validity of the present model, we compare the predicted results with the experimental data. With the average sound absorption （low frequency range） as the objective function and the fiber gaps as the design variables, an optimization method for multi-layered fibrous metals is proposed. A new fibrous layout with given porosity of multi-layered fibrous metals is suggested to achieve optimal low frequency sound absorption. The sound absorption coefficient of the optimal multi-layered fibrous metal is higher than the single- layered fibrous metal, and a significant effect of the fibrous material on sound absorption is found due to the surface Dorosity of the multi-layered fibrous.

Novel Active Learning Method for Speech Recognition

In speech recognition, acoustic modeling always requires tremendous transcribed samples, and the transcription becomes intensively time-consuming and costly. In order to aid this labor-intensive process, Active Learning （AL） is adopted for speech recognition, where only the most informative training samples are selected for manual annotation. In this paper, we propose a novel active learning method for Chinese acoustic modeling, the methods for initial training set selection based on Kullback-Leibler Divergence （KLD） and sample evaluation based on multi-level confusion networks are proposed and adopted in our active learning system, respectively. Our experiments show that our proposed method can achieve satisfying performances.

Research on Tibetan Speech Recognition Based on the Am-do Dialect

In China,Tibetan is usually divided into three major dialects:the Am-do,Khams and Lhasa dialects.The Am-do dialect evolved from ancient Tibetan and is a local variant of modern Tibetan.Although this dialect has its own specific historical and social conditions and development,there have been different degrees of communication with other ethnic groups,but all the abovementioned dialects developed from the same language:Tibetan.This paper uses the particularity of Tibetan suffixes in pronunciation and proposes a lexicon for the Am-do language,which optimizes the problems existing in previous research.Audio data of the Am-do dialect are expanded by data augmentation technology combining noise and reverberation,and the morphological characteristics and characteristics of the Tibetan language are further considered.According to the particularity of Tibetan grammar,grammatical features are used to optimize grammatical relationships and are combined with a language model,and the Am-do dialect is scored and rescored.Experimental results show that compared with the baseline,our proposed new lexicon and data augmentation technology yields a relative increase of approximately 3%in character error rates(CERs)and a relative increase of 3%-19%in the recognition rate of acoustic models and language models.

Improving the Cellular Characteristics of Aluminum Foam for Maximum Sound Absorption Coefficient Using Genetic Algorithm

Fabricating of metal foams with desired morphological parameters including pore size,porosity and pore opening is possible now using sintering technology.Thus,if it is possible to determine the morphology of metal foam to absorb sound at a given frequency,and then fabricate it through sintering,it is expected to have optimized metal foams for the best sound absorption.Theoretical sound absorption models such as Lu model describe the relationship between morphological parameters and the sound absorption coefficient.In this study,the Lu model was used to optimize the morphological parameters of aluminum metal foam for the best sound absorption coefficient.For this purpose,the Lu model was numerically solved using written codes in MATLAB software.After validating the proposed codes with benchmark data,the genetic algorithm(GA)was applied to optimize the affecting morphological parameters on the sound absorption coefficient.The optimization was carried out for the thicknesses of 5 mm to 40 mm at the sound frequency range of 250 Hz–8000 Hz.The optimized parameters ranged from 50%to 95%for porosity,0.1 mm to 4.5 mm for pore size,and 0.07 mm to 0.6 mm for pore opening size.The result of this study was applied to fabricate the desired aluminum metal foams for the best sound absorption.The novel approach applied in this study,is expected to be successfully applied in for best sound absorption in desired frequencies.

Comparison of Khasi Speech Representations with Different Spectral Features and Hidden Markov States

In this paper,we present a comparison of Khasi speech representations with four different spectral features and novel extension towards the development of Khasi speech corpora.These four features include linear predictive coding(LPC),linear prediction cepstrum coefficient(LPCC),perceptual linear prediction(PLP),and Mel frequency cepstral coefficient(MFCC).The 10-hour speech data were used for training and 3-hour data for testing.For each spectral feature,different hidden Markov model(HMM)based recognizers with variations in HMM states and different Gaussian mixture models(GMMs)were built.The performance was evaluated by using the word error rate(WER).The experimental results show that MFCC provides a better representation for Khasi speech compared with the other three spectral features.

On the certain semi-analytical models of low-frequency acoustic fields in terms of scalar-vector description

Behaviour of scalar and vector characteristics of steady-state acoustical field is modeled based on analytical-numerical approach. This field is radiated by low-frequency monochromatic point-like source in the deterministic layered shallow sea, which has various hydrologic and bottom conditions. Approach being developed is free of any mathematical approximations and without the difficulties it enables to calculate sound field vector characteristics for various