• Phonetics and Speech Sciences
• /
• v.6 no.1
• /
• pp.77-83
• /
• 2014

This paper proposes a speaker and environment clustering method in order to overcome the degradation of the speech recognition performance caused by various noise and speaker characteristics. In this paper, instead of using the distance between Gaussian mixture model (GMM) weight vectors as in the Google's approach, the distance between the adapted mean vectors based on the modified maximum a posteriori (MAP) adaptation is used as a distance measure for vector quantization (VQ) clustering. According to our experiments on the simulation data generated by adding noise to clean speech, the proposed clustering method yields error rate reduction of 10.6% compared with baseline speaker-independent (SI) model, which is slightly better performance than the Google's approach.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.25 no.12
• /
• pp.2056-2061
• /
• 2001

Non-proportional loading tests of Nylon 66 at room temperature exhibit path dependent behavior and plasticity-relaxation interactions. The uniaxial formulation of the viscoplasticity theory based on overstress (VBO), which has been used to reproduce the nonlinear strain rate sensitivity, relaxation, significant recovery and cyclic softening behaviors of Nylon 66, is extended to three-dimensions to predict the response in strain-controlled, comer-path tests. VBO consists of a flow law that is easily written for either the stress or the strain as the independent variable. The flow law depends on the overstress, the difference between the stress and the equilibrium stress that is a state variable in VBO. The evolution law of the equilibrium stress in turn contains two additional state variables, the kinematic stress and the isotropic stress. The simulations show that the constitutive model is competent at modeling the deformation behavior of Nylon 66 and other solid polymers.

• Journal of computational fluids engineering
• /
• v.6 no.4
• /
• pp.15-25
• /
• 2001

Numerical prediction of the diffusion controlled transition in a turbine gas pass is important because it can change the local heat transfer rate over a turbine blade as much as three times. In this study, the gas flow over turbine blade is simplified to the flat plate boundary layer, and an adaptive grid scheme redistributing grid points within the computation domain is proposed with a great emphasis on the construction of the grid control function. The function is sensitized to the second invariant of the mean strain tensor, its spatial gradient, and the interaction of pressure gradient and flow deformation. The transition process is assumed to be described with a κ-ε turbulence model. An elliptic solver is employed to integrate governing equations. Numerical results show that the proposed adaptive grid scheme is very effective in obtaining grid independent numerical solution with a very low grid number. It is expected that present scheme is helpful in predicting actual flow within a turbine to improve computation efficiency.

• Journal of Biomedical Engineering Research
• /
• v.24 no.1
• /
• pp.45-53
• /
• 2003

Based on the 4-compartmental pharmacokinetic model developed in PART1, target-controlled infusion(TCI) pump system was designed and evaluated. The TCI system consists of digital board including microcontroller and digital signal process(DSP), analog board, motor-driven actuator, user friendly interface, power management and controller. It provides two modes according to the drugs: plasma target concentration and effect target concentration. Anaesthetist controls the depth of anaesthesia for patients by adjusting the required concentration to maintain both plasma and effect site in drug concentration. The data estimated in DSP include infusion rate, initial load dose, and rotation number of motor encoder. During TCI operation, plasma concentration. effect site concentration, awaken concentration, context-sensitive decrement time and system error information are displayed in real time. Li-ion battery guarantees above 2 hours without power line failure. For high reliability of the system, two microprocessors were used to perform independent functions for both pharmacokinetic algorithm and motor control strategy.

• The Journal of the Acoustical Society of Korea
• /
• v.15 no.4E
• /
• pp.21-27
• /
• 1996

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.

• Proceedings of the KSPS conference
• /
• 1996.10a
• /
• pp.435-447
• /
• 1996

In this paper, we propose the method that extracts the speech feature using the hearing model through signal processing techniques. The proposed method includes the following procedure ; normalization of the short-time speech block by its maximum value, multi-resolution analysis using the discrete wavelet transformation and re-synthesize using the discrete inverse wavelet transformation, differentiation after analysis and synthesis, full wave rectification and integration. In order to verify the performance of the proposed speech feature in the speech recognition task, korean digit recognition experiments were carried out using both the DTW and the VQ-HMM. The results showed that, in the case of using DTW, the recognition rates were 99.79% and 90.33% for speaker-dependent and speaker-independent task respectively and, in the case of using VQ-HMM, the rate were 96.5% and 81.5% respectively. And it indicates that the proposed speech feature has the potential for use as a simple and efficient feature for recognition task

• Proceedings of the KSME Conference
• /
• 2000.04a
• /
• pp.132-137
• /
• 2000

In this study, model for arrest toughness is proposed in consideration of fracture behavior of composite materials. Also, the probabilistic model is proposed to describe the variability of arrest toughness due to the nonhomogeneity of material. For these models. experiments were conducted on the Carbon/Epoxy composite plates with various thickness using the impact hammer. The elastic work fatter used in J-Integral is applicable to the evaluation of energy release rate. The fracture behavior call be described by crack arrest concept and the arrest toughness is independent of the delamination size. Additionally, a probabilistic characteristics of arrest toughness is well described by the Weibull distribution function. An increasing of thickness raises a variation of arrest toughness.

• Journal of the Society of Naval Architects of Korea
• /
• v.34 no.1
• /
• pp.77-86
• /
• 1997

The advanced development in many fields of engineering and science has caused much interests and demands for crashworthiness and non-linear dynamic transient analysis of structure response. Crash and impact problems have a dominant characteristic of large deformation with material plasticity for short time scales. The structural material shows strain rate-dependent behaviors in those cases. Conventional rate-independent constitutive equations used in the general purposed finite analysis programs are inadequate for dynamic finite strain problems. In this paper, a rate-dependent constitutive equation for elastic-plastic material is developed. The plastic stretch rate is modeled based on slip model with dislocation velocity and its density so that there is neither yielding condition, nor loading conditions. Non-linear hardening rule is also introduced for finite strain. Material constants of present constitutive equation are determined by experimental data of mild steel, and the constitutive equation is applied to uniaxile tension loading.

• Journal of the Society of Naval Architects of Korea
• /
• v.51 no.6
• /
• pp.530-538
• /
• 2014

The weight estimation of floating offshore structures such as FPSO, TLP, semi-Submersibles, Floating Offshore Wind Turbines etc. in the preliminary design, is one of important measures of both construction cost and basic performance. Through both literature investigation and internet search, the weight data of floating offshore structures such as FPSO and TLP was collected. In this study, the weight estimation model was suggested for FPSO. The weight estimation model using non-linear regression analysis was established by fixing independent variables based on this data and the multiple regression analysis was introduced into the weight estimation model. Its reliability was within 4% of error rate.

• 제어로봇시스템학회:학술대회논문집
• /
• 1996.10b
• /
• pp.282-285
• /
• 1996

In the water purification plant, the raw water is promptly purified by injecting chemicals. The amount of chemicals is directly related to water quality such as turbidity, temperature, pH and alkalinity. At present, however, the process of chemical reaction to the turbidity has not been clarified as yet. Since the process of coagulant dosage has no feedback signal, the amount of chemical can not be calculated from water quality data which were sensed from the plant. Accordingly, it has to be judged and determined by Jar-Test data which were made by skilled operators. In this paper, it is concerned to model and control the coagulant dosing process using jar-test results in order to predict optimum dosage of coagulant, PAC(Polymerized Aluminium Chloride). The considering relations to the reaction of coagulation and flocculation, the five independent variables(turbidity, temperature, pH, Alkalinity of the raw water, PAC feed rate) are selected out and they are put into calculation to develope a neural network model and a fuzzy model for coagulant dosing process in water purification system. These model are utilized to predict optimum coagulant dosage which can minimize the water turbidity in flocculator. The efficacy of the proposed control schemes was examined by the field test.