The bubble size distribution is critical information to understand sound propagation and ambient noise in the ocean. To estimate the bubble size distribution in a bubbly water, the sound attenuation has been only in the conventional acoustic bubble sizing method without considering the sound speed variation. However, the effect of the sound speed variation in bubbly water cannot be neglected because of its compressibility variation. The sound attenuation is also affected by the sound speed variation. In this paper, a coherent acoustic bubble sizing inverse technique is introduced as a new bubble sizing technique with considering sound speed variation as well as the sound attenuation. This coherent sizing method is theoretically verified with the bubble distribution functions of single-size, Gaussian, and power-law functions. Its numerical test results with the coherent acoustic bubble sizing method show good agreement with the given bubble distributions.

We developed an iterative algorithm which could improve the resolution of reconstructed tomograms having random attenuation patterns and analyzed the limitation of this algorithm. The simple back-and forth propagation algorithm has depth resolution about four wavelengths. An iterative algorithm, based on back-and-forth propagation, can be used to improve the resolution of reconstructed tomograms. We analyzed the wavefield for multi-layered specimen and programmed iterative algorithm using Clanguage. Simulation results show that the images get clearer as the number of iterations increases. Also, unambiguous images can be reconstructed using this algorithm even when the layer separation is only two wavelengths. However, this iteration algorithm comes up with an incorrect solution for the number of projections less than five.

The double mode SAW resonator filters have the serious shortcoming that out-of-band rejection above the passband is quite poor. Hence a new structure of the double mode filter is designed with the COM (Coupling Of MOdes) theory with better out-of-band rejection characteristics while keeping all the beauties of the conventional type. The second goal of the design is to reduce the planar size of the structure so that mass productivity of low frequency devices can be improved. For those purposes, several IDT's and relfector gratings are added to or subtracted from the conventional structure so that the modification can lead to better overall performance as a filter. As results of the investigation, several new double mode SAW resonator filter structures are developed. A new structure shows 10dB more out-of-band rejection that the conventional type with 25% reduced device size. Good agreement is demonstrated between the analytical results and experimental results for an illustrative 325MHz four pole resonator filter on 64°YX LiNbO3.

In this paper, we propose a discriminative training algorithm for the stochastic segment model (SSM) in continuous speech recognition. As the SSM is usually trained by maximum likelihood estimation (MLE), a discriminative training algorithm is required to improve the recognition performance. Since the SSM does not assume the conditional independence of observation sequence as is done in hidden Markov models (HMMs), the search space for decoding an unknown input utterance is increased considerably. To reduce the computational complexity and starch space amount in an iterative training algorithm for discriminative SSMs, a hybrid architecture of SSMs and HMMs is programming using HMMs. Given the segment boundaries, the parameters of the SSM are discriminatively trained by the minimum error classification criterion based on a generalized probabilistic descent (GPD) method. With the discriminative training of the SSM, the word error rate is reduced by 17% compared with the MLE-trained SSM in speaker-independent continuous speech recognition.

In this paper, we propose a new evaluation methodology which can more clearly show the performance of the allophone modeling algorithm generally used in large vocabulary speech recognition. The proposed evaluation method shows the running characteristics and limitations of the modeling algorithm by testing how the variation of phonetic environments of training data affects the recognition performance and the desirable number of free parameters to be estimated. Using the method, we experiment results, we conclude that, in vocabulary-independent recognition task, the phonetic diversity of training data greatly affects the robustness of model, and it is necessary to develop a proper measure which can determine the number of states compromizing the robustness and the precision of the HMnet better than the conventional modeling efficiency.

To get accurate vibration analysis of rotor-bearing systems, finite element models of high speed rotating shaft, unbalance disk, and fluid film journal bearing are developed. The study includes the effects of rotary inertia, gyroscopic moment, damping, shear deformation, and axial torque in the same model. It does not include the axial force effect, but the extension is straighforward. The finite elements developed can be used in the analysis design of any type of multiple rotor bearing system. To show the accuracy of the models, numerical examples are demonstrated.

One common procedure in the approximate solution of a nonclassically damped linear system is to neglect the off-diagonal elements of the normalized damping matrix. A tight error bound, which can be computed with relative ease, is given for this method of solution. The role that modal coupling plays in the control of error is clarified. If the normalized damping matrix is strongly diagonally dominant, it is shown that adequate frequency separation is not necessary to ensure small errors.

An extension of the well-known Fourier transform is developed in this paper. It is denoted as the generalized Fourier transform(GFT), since it encompasses the Fourier transform as its special case. The first idea of this extension can be found on [1]. In the definition of the N-point discrete GFT, it first construct a passband in time which functions as a window in the time domain. An appropriate interpretation of each variables are introduced during the definition of the GFT, followed by the formal derivation of the inverse GFT. This transform pair is similar to the windowing in the frequency domain such as the subband coding technique (or filter bank approach) and could be extended to the wavelet transform.

The aims of this study were to investigate the covariation between the extent of activity disturbances and general annoyance, and to study the relation between the extent of annoyance reactions and noise sensitivity. This paper presents a description of a model developed by taking into account self-rated noise sensitivity and noise rating(i,e., annoyance) for road traffic. The results indicate that there are large variations in noise sensitivity which is independent on the level of noise. It is also found that the extent of all activity disturbances decreased with decreased general annoyance. The paper suggests a normative approach to predicting individual's reaction to noise exposure, based on a periodically observed relationship between the prevalence of activity disturbance and annoyance. The model also predicts a road traffic-noise adaptatin level for each individual.

In shallow water with a thermocline, the characteristics of sound propagation strongly depend on the signal frequency. When only one of the source and the receiver is above the themocline, it is known that the intensity of the received signal changes largely and almost periodically as the signal frequency varies. This is the so-called channel resonance. By using the ray-mode approach, the formula relating the resonance frequency and the sound speed profile is obtained, and the resonance phenomenon is analyzed. Also this analysis is verified by numerical calculation.

A relatively simple theoretical model for predicting the shock thickness is applied to the weak shock propagation through sea water, where the boric acid and the magnesium sulfate are the major relaxation processes. The relaxation effects increases the shock thickness by the factor of 103 compared with the thickness based on the classical absorption only. In seawater with the ambient pressure of 125 atm and 15℃ temperature, the effects of the boric acid are dominant when the peak pressure is less than 0.3 atm and 3 atm. For the shocks of peak pressure greater than 5 atm, the effects of the classical absorption theory is enough to describe the shock thickness. The effects of the ambient pressure and temperature on the shock thickness are also evaluated.