Showing papers by "Nagoya Institute of Technology published in 2007"

Statistical Parametric Speech Synthesis

Abstract: This paper gives a general overview of techniques in statistical parametric speech synthesis. One of the instances of these techniques, called HMM-based generation synthesis (or simply HMM-based synthesis), has recently been shown to be very effective in generating acceptable speech synthesis. This paper also contrasts these techniques with the more conventional unit selection technology that has dominated speech synthesis over the last ten years. Advantages and disadvantages of statistical parametric synthesis are highlighted as well as identifying where we expect the key developments to appear in the immediate future.

Voice Conversion Based on Maximum-Likelihood Estimation of Spectral Parameter Trajectory

Abstract: In this paper, we describe a novel spectral conversion method for voice conversion (VC). A Gaussian mixture model (GMM) of the joint probability density of source and target features is employed for performing spectral conversion between speakers. The conventional method converts spectral parameters frame by frame based on the minimum mean square error. Although it is reasonably effective, the deterioration of speech quality is caused by some problems: 1) appropriate spectral movements are not always caused by the frame-based conversion process, and 2) the converted spectra are excessively smoothed by statistical modeling. In order to address those problems, we propose a conversion method based on the maximum-likelihood estimation of a spectral parameter trajectory. Not only static but also dynamic feature statistics are used for realizing the appropriate converted spectrum sequence. Moreover, the oversmoothing effect is alleviated by considering a global variance feature of the converted spectra. Experimental results indicate that the performance of VC can be dramatically improved by the proposed method in view of both speech quality and conversion accuracy for speaker individuality.

Solvothermal Synthesis of Multiple Shapes of Silver Nanoparticles and Their SERS Properties

Abstract: The controllable synthesis of nanocrystals with different shapes is very important and challenging. In the present work, silver nanoparticles with different structural architectures, from nanorods, triangular plates, hexagonal plates, and nanocubes to polyhedrons have been synthesized successfully in high yield by a solvothermal process. Especially, the unique silver enneahedral nanoplates are also observed. These nanoparticles exhibit tunable surface plasmon resonance properties from the visible to near-infrared regions. Those nanoparticles were also self-assembled on glass substrates and evaluated as potential surface-enhanced Raman scattering (SERS) substrates using trans-1,2-bis(4-pyridyl)ethylene molecules. Thanks to the enhanced local field effect around their sharp corners and edges, those Ag triangular plates exhibit enhanced SERS properties and can serve as high-sensitivity substrates for SERS-based measurements.

Details of the Nitech HMM-Based Speech Synthesis System for the Blizzard Challenge 2005

Abstract: In January 2005, an open evaluation of corpus-based text-to-speech synthesis systems using common speech datasets, named Blizzard Challenge 2005, was conducted. Nitech group participated in this challenge, entering an HMM-based speech synthesis system called Nitech-HTS 2005. This paper describes the technical details, building processes, and performance of our system. We first give an overview of the basic HMM-based speech synthesis system, and then describe new features integrated into Nitech-HTS 2005 such as STRAIGHT-based vocoding, HSMM-based acoustic modeling, and a speech parameter generation algorithm considering GV. Constructed Nitech-HTS 2005 voices can generate speech waveforms at 0.3 ×RT (real-time ratio) on a 1.6 GHz Pentium 4 machine, and footprints of these voices are less than 2 Mbytes. Subjective listening tests showed that the naturalness and intelligibility of the Nitech-HTS 2005 voices were much better than expected.

Extracting influential nodes for information diffusion on a social network

Abstract: We consider the combinatorial optimization problem of finding the most influential nodes on a large-scale social network for two widely-used fundamental stochastic diffusion models. It was shown that a natural greedy strategy can give a good approximate solution to this optimization problem. However, a conventional method under the greedy algorithm needs a large amount of computation, since it estimates the marginal gains for the expected number of nodes influenced by a set of nodes by simulating the random process of each model many times. In this paper, we propose a method of efficiently estimating all those quantities on the basis of bond percolation and graph theory, and apply it to approximately solving the optimization problem under the greedy algorithm. Using real-world large-scale networks including blog networks, we experimentally demonstrate that the proposed method can outperform the conventional method, and achieve a large reduction in computational cost.

Multi-issue negotiation protocol for agents: exploring nonlinear utility spaces

Abstract: Multi-issue negotiation protocols have been studied widely and represent a promising field since most negotiation problems in the real world involve interdependent multiple issues The vast majority of this work has assumed that negotiation issues are independent, so agents can aggregate the utilities of the issue values by simple summation, producing linear utility functions In the real world, however, such aggregations are often unrealistic We cannot, for example, just add up the value of car's carburetor and the value of car's engine when engineers negotiate over the design a car These value of these choices are interdependent, resulting in nonlinear utility functions In this paper, we address this important gap in current negotiation techniques We propose a negotiation protocol where agents employ adjusted sampling to generate proposals, and a bidding-based mechanism is used to find social-welfare maximizing deals Our experimental results show that our method substantially outperforms existing methods in large non-linear utility spaces like those found in real world contexts

Cinchona Alkaloid-Catalyzed Enantioselective Monofluoromethylation Reaction Based on Fluorobis(phenylsulfonyl)methane Chemistry Combined with a Mannich-type Reaction

Abstract: A catalytic protocol for the unprecedented enantioselective monofluoromethylation of in situ generated prochiral imines using 1-fluorobis(phenylsulfonyl)methane (FBSM, 1) in the presence of a chiral PTC based on the Mannich-type reaction followed by reductive desulfonylation has been developed.

Development of Ruthenium−Hydroxyapatite-Encapsulated Superparamagnetic γ-Fe2O3 Nanocrystallites as an Efficient Oxidation Catalyst by Molecular Oxygen

Abstract: Magnetic γ-Fe2O3 nanocrystallites dispersed in a hydroxyapatite matrix (HAP-γ-Fe2O3) have been synthesized for the first time as a new catalyst support. The cation-exchange ability of the external HAP surface enables an equimolar substitution of Ru for Ca to form a catalytically active center (RuHAP-γ-Fe2O3). Characterization by several spectroscopic methods demonstrated the formation of γ-Fe2O3 nanocrystallites within the HAP matrix having a mean diameter of 8.0 nm (σ = 18.4 A, σ/d = 21.7%). The Ru species was proven to exist in the monomeric form in a highly dispersed manner. The RuHAP-γ-Fe2O3 exhibits superior catalytic activity for the oxidation of various alcohols to the corresponding carbonyl compounds using molecular oxygen as a primary oxidant. The magnetic properties of the RuHAP-γ-Fe2O3 provided a convenient route for separation of the catalyst from the reaction mixture by application of an external permanent magnet. No metal leaching was detected, and the spent catalyst could be recycled withou...

Self-assembled silver nanochains for surface-enhanced Raman scattering.

Abstract: Surface-enhanced Raman scattering (SERS) integrates high levels of sensitivity with spectroscopic precision and has tremendous potential for chemical and biomolecular sensing. The key to the wider application of Raman spectroscopy using roughened metallic surfaces is the development of highly enhancing substrates for analytical purposes. Here, we demonstrate a simple strategy for self-assembling silver nanochains on glass substrates for sensitive SERS substrates. The chain length of short Ag nanochains can be controlled by adjusting the concentration of cetyltrimethylammonium bromide (CTAB) and 11-mercaptoundecanoic acid (MUA). CTAB with appropriate concentration serves as the “glue” that can link the {100} facets of two neighboring Ag nanoparticles. MUA is found to be effective in “freezing up” the aggregation of Ag short chains and preventing them from further aggregating into a long chainlike network structure. The surface plasmon bands can be tuned over an extended wavelength range by controlling the ...

A density‐dependent elastoplastic hydro‐mechanical model for unsaturated compacted soils

Abstract: This paper presents a three-dimensional elastoplastic constitutive model for predicting the hydraulic and mechanical behaviour of unsaturated soils It is based on experimental results obtained from a series of controlled-suction triaxial tests on unsaturated compacted clay with different initial densities Hydraulic hysteresis in the water-retention behaviour is modelled as an elastoplastic process, with the elastic part modelled by a series of scanning curves and the elastoplastic part modelled by the main drying and wetting curves The effect of void ratio on the water-retention behaviour is studied using data obtained from controlled-suction wetting–drying cyclic tests on unsaturated compacted clay with different initial densities The effect of the degree of saturation on the stress–strain-strength behaviour and the effect of void ratio on the water-retention behaviour are considered in the model, as is the effect of suction on the hydraulic and mechanical behaviour The initial density dependency of the compacted soil behaviour is modelled by experimental relationships between the initial density and the corresponding yield stress and, thereby, between the initial density and the normal compression line The model is generalized to three-dimensional stress states by assuming that the shapes of the failure and yield surfaces in the deviatoric stress plane are given by the Matsuoka–Nakai criterion Model predictions of the stress–strain and water-retention behaviour are compared with those obtained from triaxial tests with different initial densities under isotropic compression, triaxial compression and triaxial extension, with or without variation in suction The comparisons indicate that the model accurately predicts the hydraulic and mechanical behaviour of unsaturated compacted soils with different initial densities using the same material constant Copyright © 2006 John Wiley & Sons, Ltd

Experimental study on miscible viscous fingering involving viscosity changes induced by variations in chemical species concentrations due to chemical reactions

Abstract: When a reactive and miscible less-viscous liquid displaces a more-viscous liquid in a Hele-Shaw cell, reactive miscible viscous fingering takes place. We succeed in showing experimentally how a reactive miscible viscous fingering pattern in a radial Hele-Shaw cell changes when the viscosity of the more-viscous liquid is varied owing to variation in chemical species concentration induced by an instantaneous chemical reaction. This is done by making use of a polymer solution's dependence of viscosity on pH. When the viscosity is increased by the chemical reaction, the shielding effect is suppressed and the fingers are widened. As a result, the ratio of the area occupied by the fingering pattern in a circle whose radius is the length of the longest finger is larger in the reactive case than in the non-reactive case. When the viscosity is decreased by the chemical reaction, in contrast, the shielding effect is enhanced and the fingers are narrowed. These lead to the area ratio being smaller in the reactive case than in the non-reactive case. A physical model to explain this change in the fingering pattern caused by the chemical reaction is proposed.

Study of the Size Effect on Friction Conditions in Microextrusion—Part I: Microextrusion Experiments and Analysis

Abstract: Microforming is a relatively new realm of manufacturing technology that addresses the issues involved in the fabrication of metallic microparts, i.e., metallic parts that have at least two characteristic dimensions in the sub-millimeter range. The recent trend towards miniaturization of products and technology has produced a strong demand for such metallic microparts with extremely small geometric features and high tolerances. Conventional forming technologies, such as extrusion, have encountered new challenges at the microscale due to the influence of "size effects" that tend to be predominant at this length scale. One of the factors that of interest is friction. The two companion papers investigate the frictional behavior and size effects observed during microextrusion in Part I and in a stored-energy Kolsky bar test in Part II. In this first paper, a novel experimental setup consisting of forming assembly and a loading stage has been developed to obtain the force-displacement response for the extrusion of pins made of brass (Cu/Zn: 70/30). This experimental setup is used to extrude pins with a circular cross section that have a final extruded diameter ranging from 1.33 mm down to 570 μm. The experimental results are then compared to finite-element simulations and analytical models to quantify the frictional behavior. It was found that the friction condition was nonuniform and showed a dependence on the dimensions (or size) of the micropin under the assumption of a homogeneous material deformation. Such assumption will be eliminated in Part II where the friction coefficient is more directly measured. Part I also investigates the validity of using high-strength/low-friction die coatings to improve the tribological characteristics observed in micro-extrusion. Three different extrusion dies coated with diamondlike carbon with silicon (DLC-Si), chromium nitride (CrN), and titanium nitride (TiN) were used in the microextrusion experiments. All the coatings worked satisfactorily in reducing the friction and, correspondingly, the extrusion force with the DLC-Si coating producing the best results.

Fourth-Moment Standardization for Structural Reliability Assessment

Abstract: In structural reliability analysis, the uncertainties related to resistance and load are generally expressed as random variables that have known cumulative distribution functions. However, in practical applications, the cumulative distribution functions of some random variables may be unknown, and the probabilistic characteristics of these variables may be expressed using only statistical moments. In the present paper, in order to conduct structural reliability analysis without the exclusion of random variables having unknown distributions, the third-order polynomial normal transformation technique using the first four central moments is investigated, and an explicit fourth-moment standardization function is proposed. Using the proposed method, the normal transformation for indepen- dent random variables with unknown cumulative distribution functions can be realized without using the Rosenblatt transformation or Nataf transformation. Through the numerical examples presented, the proposed method is found to be sufficiently accurate in its inclusion of the independent random variables which have unknown cumulative distribution functions, in structural reliability analyses with minimal additional computational effort.

An Excitation Model for HMM-Based Speech Synthesis Based on Residual Modeling

Abstract: This paper describes a trainable excitation approach to eliminate the unnaturalness of HMM-based speech synthesizers. During the waveform generation part, mixed excitation is constructed by state-dependent filtering of pulse trains and white noise sequences. In the training part, filters and pulse trains are jointly optimized through a procedure which resembles analysis-bysynthesis speech coding algorithms, where likelihood maximization of residual signals (derived from the same database which is used to train the HMM-based synthesizer) is pursued. Preliminary results show that the novel excitation model in question eliminates the unnaturalness of synthesized speech, being comparable in quality to the the best approaches thus far reported to eradicate the buzziness of HMM-based synthesizers.

Characterization of Electrode/Electrolyte Interface with X-Ray Reflectometry and Epitaxial-Film LiMn2O4 Electrode

Abstract: Structural changes at electrode/electrolyte interface of a lithium cell were studied by X-ray reflectometry and two-dimensional model electrodes with a restricted lattice plane of LiMn 2 O 4 . The electrodes were constructed with an epitaxial film synthesized by the pulsed laser deposition method. The orientation of the film depends on the substrate plane; the (111), (110), and (100) planes of LiMn 2 O 4 grew on the (111), (110), and (100) planes of the SrTiO 3 substrates, respectively. The ex situ reflectometry indicated that a thin impurity layer covered the lattice plane of the as-grown film. The impurity layer was dissolved and a solid-electrolyte-interface-like phase appeared after the electrode was soaked into the electrolyte. A defect layer was formed in the (111) plane, whereas no density changes were detected for the other lattice planes. The in situ observation clarified that the surface reactivity depended on the lattice planes of the spinel; the defect layer at the (111) plane was stable during the electrochemical reaction, whereas a slight decrease in the film thickness was observed for the (110) plane. Our surface characterization of the intercalation electrode indicated that the surface structure changes during the pristine stage of the change-discharge processes and these changes are dependent on the lattice orientation of LiMn 2 O 4 .

Mechanical Characterization of Brain Tissue in High-Rate Compression

Abstract: Mechanical properties of brain tissue in high strain region are indispensable for the analysis of brain damage during traffic accidents. However, accurate data on the mechanical behavior of brain tissue under impact loading condition are sparse. In this study, mechanical properties of porcine brain tissues were characterized in their cylindrical samples cored out from their surface. The samples were compressed in their axial direction at strain rates ranging from 1 to 50 s-1. Stress relaxation test was also conducted following rapid compression with a rise time of ∼30 ms to different strain levels (20-70%). Brain tissue exhibited stiffer responses under higher impact rates: initial elastic modulus was 5.7±1.6, 11.9±3.3, 23.8±10.5 kPa (mean±SD) for strain rate of 1, 10, 50 s-1, respectively. We found that stress relaxation K(t,e) could be analysed in time and strain domains separately. The relaxation response could be expressed as the product of two mutually independent functions of time and strain as: K(t,e)=G(t)σe(e), where σe(e) is an elastic response, i.e., the peak stress in response to a step input of strain e, and G(t) is a reduced relaxation function: G(t)=0.642e-t/0.0207+0.142e-t/0.482+0.216e-t/18.9, i.e., the time-dependent stress response normalized by the peak stress. The reduced relaxation function obtained here will serve as a useful tool to predict mechanical behavior of brain tissue in compression with strain rate greater than 10 s-1.

Effect of (Na,K)-Excess Precursor Solutions on Alkoxy-Derived (Na,K)NbO3 Powders and Thin Films

Abstract: Lead-free (Na05K05)NbO3 (NKN) powders and thin films were fabricated from stoichiometric (Na/K=50/50), 4 mol % excess (52/52, 53/51, and 55/49), and 10 mol % excess (55/55, 56/54, 58/52, and 60/50) precursor solutions by the sol–gel process The NKN55/55 powder heat-treated at 800 °C showed an orthorhombic phase and rectangular grains whose size was estimated to be about 300 nm With increasing Na/K ratio, grain size increased to about 1 µm and the cubic–tetragonal (Tc) phase transition peak emerged at 392 °C in a differential thermal analysis (DTA) curve On the other hand, single-phase NKN thin films were fabricated at 600 °C from the (Na,K)-excess precursor solutions (58/52 and 60/50) In particular, the NKN58/52 thin film showed a low leakage current density (10-7 A/cm2 at 40 kV/cm), and the maximum polarization Pmax and dielectric constant er at 1 kHz were 91 µC/cm2 and 725, respectively

Growth behavior of compound layers during reactive diffusion between solid Cu and liquid Al

Abstract: In order to examine experimentally the kinetics of the reactive diffusion between solid Cu and liquid Al, Cu/Al diffusion couples were isothermally annealed in the temperature range between T = 973 and 1073 K. Owing to annealing, compound layers consisting of the β, γ1 and ɛ2 phases are produced between the Cu-rich solid (α) and Al-rich liquid (L) phases. The β, γ1 and ɛ2 phases are the only stable compounds at T = 973–1073 K. The mean thickness of each compound layer is proportional to a power function of the annealing time. For the β layer, the exponent of the power function is close to 0.5 at T = 1023–1073 K, but nearly equal to 0.25 at T = 973 K. On the other hand, for the γ1 layer, the exponent takes values between 0.25 and 0.5 at T = 1023–1073 K, but that smaller than 0.25 at T = 973 K. The exponent smaller than 0.5 indicates that grain boundary diffusion predominantly controls the growth of the compound layer and grain growth occurs at a certain rate. In contrast, the ɛ2/L interface migrates towards the α phase. The migration distance of the ɛ2/L interface is much greater than the total thickness of the compound layers, and proportional to the square root of the annealing time. Consequently, the migration of the ɛ2/L interface is governed by interdiffusion in the L phase. According to estimation, the interdiffusion coefficient is much greater for the L phase than for the solid phases. As a result, the ɛ2/L interface migrates towards the α phase, and the migration rate of the interface is much greater than the overall growth rate of the compound layers.

Dominant factors influencing whole-body average SAR due to far-field exposure in whole-body resonance frequency and GHz regions.

Abstract: Electromagnetic (EM) absorption of the human body for far-field exposure at the International Commission on Non-Ionizing Radiation Protection (ICNIRP) reference level has two peaks in the resonance frequency and GHz regions. Dominant factors influencing whole-body average specific absorption rate (SAR) in these two frequency regions have not yet been revealed sufficiently. The main purpose of this study is to clarify the dominant factors influencing EM power absorption in terms of whole-body average SAR in an anatomically based model compared with those in a homogeneous anthropomorphic model and corresponding cuboid models. Computational results show that EM absorption peak in the resonance frequency region greatly depends on the electric properties of tissue, while the peak in the GHz region is affected mainly by the surface area of the model. Bioelectromagnetics 28:484–487, 2007. © 2007 Wiley-Liss, Inc.

Controlling the aggregation behavior of gold nanoparticles

Abstract: The aggregation of Au nanoparticles (NPs) in solution is influenced by cationic and oligocationic species. The polarization of the conduction electron oscillations in adjacent gold nanoparticles causes a new red-shifted plasmon absorbance attributed to the coupling of the plasmon absorbance of the particles. This appearance of an additional plasmon band is of particular interest to the field of SERS and has led to research works directed at the stabilization of small colloid aggregates in solution. The surface plasmon coupling can be tuned by controlling the aggregation of gold nanoparticles by the addition of some “cross-linking” agent. Here we develop a simple method to fabricate linear-chainlike aggregates of gold nanoparticles (so-called nanochains), tuning the linear optical properties in a wide wavelength range from visible to the near-infrared. The aggregation behavior and linear self-assembly mechanism of citrate-stabilized gold colloids as provoked by the addition of cetyltrimethylammonium bromide (CTAB) and 11-mercaptoundecanoic acid (MUA) are also analyzed. The line-assembly mechanism of gold nanochain is attributed to the preferential binding of CTAB molecules on a certain facet of gold NPs and the Au NP electrostatic interactions. We also found that the 11-mercaptoundecanoic acid was effective to prevent the further aggregation of CTAB-modified gold colloids.