Showing papers by "Nagoya Institute of Technology published in 2000"

Speech parameter generation algorithms for HMM-based speech synthesis

Abstract: This paper derives a speech parameter generation algorithm for HMM-based speech synthesis, in which the speech parameter sequence is generated from HMMs whose observation vector consists of a spectral parameter vector and its dynamic feature vectors. In the algorithm, we assume that the state sequence (state and mixture sequence for the multi-mixture case) or a part of the state sequence is unobservable (i.e., hidden or latent). As a result, the algorithm iterates the forward-backward algorithm and the parameter generation algorithm for the case where the state sequence is given. Experimental results show that by using the algorithm, we can reproduce clear formant structure from multi-mixture HMMs as compared with that produced from single-mixture HMMs.

Extended pseudo-Voigt function for approximating the Voigt profile

Abstract: The formula of the pseudo-Voigt function expressed by a weighted sum of Gaussian and Lorentzian functions is extended by adding two other types of peak functions in order to improve the accuracy when approximating the Voigt profile. The full width at half-maximum (FWHM) values and mixing parameters of the Gaussian, the Lorentzian and the other two component functions in the extended formula can be approximated by polynomials of a parameter ρ = ΓL/(ΓG + ΓL), where ΓG and ΓL are the FWHM values of the deconvoluted Gaussian and Lorentzian functions, respectively. The maximum deviation of the extended pseudo-Voigt function from the Voigt profile is within 0.12% relative to the peak height when sixth-order polynomial expansions are used. The systematic errors of the integrated intensity ΓG and ΓL, estimated by fitting the extended formula to Voigt profiles, are typically less than 1/10 of the errors arising from the application of the original formula of the pseudo-Voigt approximation proposed by Thompson et al. [J. Appl. Cryst. (1987), 20, 79–83], while the time required for computation of the extended formula is only about 2.5 relative to the computation time required for the original formula.

New Point Estimates for Probability Moments

Abstract: There are many areas of structural safety and structural dynamics in which it is often desirable to compute the first few statistical moments of a function of random variables. The usual approximation is by the Taylor expansion method. This approach requires the computation of derivatives. In order to avoid the computation of derivatives, point estimates for probability moments have been proposed. However, the accuracy is quite low, and sometimes, the estimating points may be outside the region in which the random variable is defined. In the present paper, new point estimates for probability moments are proposed, in which increasing the number of estimating points is easier because the estimating points are independent of the random variable in its original space and the use of high-order moments of the random variables is not required. By using this approximation, the practicability and accuracy of point estimates can be much improved.

Speaker interpolation for HMM-based speech synthesis system.

Abstract: This paper describes an approach to voice characteristics conversion for an HMM-based text-to-speech synthesis system using speaker interpolation.Although most text-to-speech synthesis systems which synthesize speech by concatenating speech units can synthesize speech with acceptable quality, they still cannot synthesize speech with various voice quality such as speaker individualities and emotions;In order to control speaker individualities and emotions, therefore, they need a large database, which records speech units with various voice characteristics in sythesis phase.On the other hand, our system synthesize speech with untrained speaker’s voice quality by interpolating HMM parameters among some representative speakers’ HMM sets.Accordingly, our system can synthesize speech with various voice quality without large database in synthesis phase.An HMM interpolation technique is derived from a probabilistic similarity measure for HMMs, and used to synthesize speech with untrained speaker’s voice quality by interpolating HMM parameters among some representative speakers’ HMM sets.The results of subjective experiments show that we can gradually change the voice quality of synthesized speech from one’s to the other’s by changing the interpolation ratio.

Porous Titania Ceramic Prepared by Mimicking Silicified Wood

Abstract: A porous titania ceramic with a woodlike microstructure that was analogous to silicified wood was prepared. The production of “titanified wood” was performed using the following process: (i) introduction of titanium tetraisopropoxide into wood materials via vacuum infiltration, (ii) hydrolysis of the titanium tetraisopropoxide in the cell structure to form a titania gel, and (iii) firing at a temperature of 600°–1400°C in air. The resulting titania ceramic had the same external and internal forms of the original wood.

Randomized initialization protocols for ad hoc networks

Abstract: Ad hoc networks are self-organizing entities that are deployed on demand in support of various events including collaborative computing, multimedia classroom, disaster-relief, search-and-rescue, interactive mission planning, and law enforcement operations. One of the fundamental tasks that have to be addressed when setting up an ad hoc network (AHN, for short) is initialization. This involves assigning each of the n stations in the AHN a distinct ID number (e.g., a local IP address) in the range from 1 to n. Our main contribution is to propose efficient randomized initialization protocols for AHNs. We begin by showing that if the number n of stations is known beforehand, an n-station, single-channel AHN can be initialized with probability exceeding 1-(1/n), in en+O(/spl radic/(nlogn)) time slots, regardless of whether the AHN has collision detection capability. We then go on to show that even if n is not known in advance, an n-station, single-channel AHN with collision detection can be initialized with probability exceeding 1-(1/n), in (10n)/3+O(/spl radic/(n 1n n)) time slots. Using this protocol as a stepping stone, we then present an initialization protocol for the n-station, k-channel AHN with collision detection that terminates with probability exceeding 1-(1/n), in (10n)/(3k)+O(/spl radic/(n 1n n)/k) time slots. Finally, we look at the case where the collision detection capability is not present. Our first result in this direction is to show that the task of electing a leader in an n-station, single-channel AHN can be completed with probability exceeding 1-(1/n), in fewer than 11.37(iog n)/sup 2/+2.39 log n time slots. This leader election protocol allows us to design an initialization protocol for the n-station, single-channel AHN with no collision detection that terminates with probability exceeding 1-(1/n), in fewer than 5.67n+O(/spl radic/(n 1n n)) time slots, even if n is not known beforehand. We then discuss an initialization protocol for the n-station, k-channel AHN with no collision detection that terminates with probability exceeding 1-(1/n), in fewer than 5.67(n/k)+O(/spl radic/(n 1n n)/k) time slots, whenever k/spl les/n/((log n)/sup 3/).

Pseudogap Formation in the Intermetallic Compounds ( Fe 1 − x V x ) 3 Al

Abstract: Optical conductivity data of the intermetallic compounds $({\mathrm{Fe}}_{1\ensuremath{-}x}{\mathrm{V}}_{x}{)}_{3}\mathrm{Al}$ ( $0\ensuremath{\le}x\ensuremath{\le}0.33$) reveal that their density of states around the Fermi energy ( ${E}_{F}$) is strongly reduced as $x$ is increased. In particular, ${\mathrm{Fe}}_{2}\mathrm{VAl}$ ( $x\phantom{\rule{0ex}{0ex}}=\phantom{\rule{0ex}{0ex}}0.33$) has a deep, well-developed pseudogap of $0.1--0.2\mathrm{eV}$ at ${E}_{F}$ and a small density ( $\ensuremath{\sim}5\ifmmode\times\else\texttimes\fi{}{10}^{20}\phantom{\rule{0ex}{0ex}}{\mathrm{cm}}^{\ensuremath{-}3}$) of carriers, which is highly unusual for intermetallic compounds. It is shown that the pseudogap results from the band structure of ${\mathrm{Fe}}_{2}\mathrm{VAl}$, rather than from temperature-dependent correlation effects. Based on the present results, we propose a simple model that consistently explains both the semiconductorlike transport and the metallic photoemission results previously observed for ${\mathrm{Fe}}_{2}\mathrm{VAl}$.

Enantioselective Reactions of Configurationally Unstable α‐Thiobenzyllithium Compounds

Abstract: Trotz ihrer Konfigurationslabilitat, die sie fur eine asymmetrische Deprotonierung ungeeignet macht, konnen α-Sulfenylcarbanionen enantioselektive Substitutionsreaktionen mit hoher Stereoselektivitat eingehen [Gl. (1)]. Der Schlussel zur erfolgreichen asymmetrischen Induktion ist die dynamische kinetische Racematspaltung des Komplexes aus der Organolithiumverbindung und einem chiralen Liganden, vorzugsweise einem Bisoxazolinderivat.

Polycation liposomes, a novel nonviral gene transfer system, constructed from cetylated polyethylenimine

Abstract: A novel gene transfer system was developed by using liposomes modified with cetylated polyethylenimine (PEI, MW 600). This polycation liposome, PCL, showed remarkable transfection efficiency as monitored by the expression of the GFP reporter gene. Most conventional cationic liposomes require phosphatidylethanolamine or cholesterol as a component, although PCLs did not. Egg yolk phosphatidylcholine- and dipalmitoylphosphatidylcholine-based PCL were as effective as dioleoylphosphatidylethanolamine-based PCLs for gene transfer. Concerning the cytotoxicity against COS-1 cells and hemolytic activity, the PCL was superior to conventional cationic liposome preparations. Furthermore, the transfection efficacy of PCLs was enhanced, instead of being diminished, in the presence of serum. Effective gene transfer was observed in all eight malignant and two normal cells line tested, as well as in COS-1 cells. We also examined the effect of the molecular weight of PEI on PCL-mediated gene transfer, and observed that PEI with a MW of 1800 Da was as effective as that with one of 600, but that PEI of 25 000 was far less effective. Finally, an in vivo study was done in which GFP was effectively expressed in mouse liver after injection of PCL via the portal vein. Thus, PCL represents a new system useful for transfection and gene therapy.

Vitamin E supplementation attenuates leakage of enzymes following 6 successive days of running training.

Abstract: The purpose of this study was to examine whether vitamin E supplementation in humans would attenuate an increase of serum enzymes as an indirect marker of muscle damage following a sudden large increase in the running distance in a 6-day running training or not. A randomized and placebo-controlled study was carried out on fourteen male runners who were supplied vitamin E (alpha-tocopherol 1200 IU x day(-1); E) or placebo (P) 4 weeks prior to (T1) and during 6 successive days of running training (48.3 +/- 5.7 km x day(-1), means +/- SD). Resting venous blood samples were obtained before maximal treadmill running, at T1, the day immediately before (T2), the next day (T3), and three weeks (T4) after the running training. Serum levels of alpha-tocopherol, lipid peroxidation products (thiobarbituric acid; TBA), creatine kinase (CK), lactate dehydrogenase (LDH), and LDH isozyme 1-5 were quantitatively analyzed. No significant difference was found in maximal oxygen uptake (VO2max) and maximal heart rates following the exhaustive exercise between the P and E group during the experiments. Vitamin E supplementation significantly increased serum alpha-tocopherol (p<0.001) and decreased TBA levels (p < 0.001) compared with pre-supplementation levels. Although serum CK and LDH activities increased significantly at T3 in either group, significantly lower CK (p < 0.05) and LDH (p < 0.001) levels were observed in the E group compared with the P group. The ratio of LDH1 to LDH2 (LDH1/LDH2) decreased significantly at T3 in either group compared with the T1 levels, since there was no significant difference in the LDH1/LDH2 between the P and E group throughout the experiments. These results indicate that vitamin E supplementation can reduce the leakage of CK and LDH following 6 successive days of endurance running. The protective effect of vitamin E against free radicals probably inhibits free-radical-induced muscle damage caused by a sudden large increase in the running distance.

Improved characteristics of InGaN multiple-quantum-well light-emitting diode by GaN/AlGaN distributed Bragg reflector grown on sapphire

Abstract: An InGaN multiple-quantum-well light-emitting diode (LED) containing a GaN/AlGaN distributed Bragg reflector has been grown on a sapphire substrate by metalorganic chemical vapor deposition. Comparing with the conventional LED, the output power has been improved from 79 to 120 μW under 20 mA direct current biasing condition and the external quantum efficiency has been also improved from 0.16% to 0.23% under 10 mA dc current.

Highly Enantioselective Reactions of Configurationally Labile α-Thioorganolithiums Using Chiral Bis(oxazoline)s via Two Different Enantiodetermining Steps

Abstract: A cumene solution of α-stannyl benzyl phenyl sulfide was treated with n-BuLi and bis(oxazoline)-iPr at −78 °C and subsequently with benzophenone to give the product with 99% ee. We confirmed that the reaction of α-lithio benzyl phenyl sulfide proceeds through a dynamic kinetic resolution pathway. The enantioselective reactions of α-lithio benzyl 2-pyridyl sulfide gave the products with stereochemistry reverse to that obtained in the reaction of benzyl phenyl sulfide. We confirmed that this reaction proceeds through a dynamic thermodynamic resolution pathway in which the reaction with an electrophile proceeds faster than interconversion between the diastereomeric complexes.

On stability analysis of 2-D systems based on 2-D Lyapunov matrix inequalities

Abstract: A sufficient condition for the stability of two-dimensional (2-D) systems described in the Fornasin-Marchesini model is presented in a succinct form. The condition is expressed as a Lyapunov-like matrix inequality involved with the parallel addition of two positive definite matrices.

Recessed gate AlGaN/GaN modulation-doped field-effect transistors on sapphire

Abstract: A recessed gate AlGaN/GaN modulation-doped field-effect transistor (MODFET) has been grown on a sapphire substrate by metalorganic chemical vapor deposition. The two-dimensional electron gas mobility as high as 9260 cm2/V s with the sheet carrier density 4.8×1012 cm−2 was measured at 4.6 K for the AlGaN/GaN heterostructure on the sapphire substrate. The recessed gate device showed the maximum extrinsic transconductance 146 mS/mm and drain-source current 900 mA/mm for the AlGaN/GaN MODFET with a gate length 2.1 μm at 25 °C. At an elevated temperature of 350 °C, the maximum extrinsic transconductance and drain-source current were 62 mS/mm and 347 mA/mm, respectively.

Third-Moment Standardization for Structural Reliability Analysis

Abstract: First- and second-order reliability methods are generally considered to be among the most useful for computing structural reliability. In these methods, the uncertainties included in resistances and loads are generally expressed as continuous random variables that have a known cumulative distribution function. The Rosenblatt transformation is a fundamental requirement for structural reliability analysis. However, in practical applications, the cumulative distribution functions of some random variables are unknown, and the probabilistic characteristics of these variables may be expressed using only statistical moments. In the present study, a structural reliability analysis method with inclusion of random variables with unknown cumulative distribution functions is suggested. Normal transformation methods that make use of high-order moments are investigated, and an accurate third-moment standardization function is proposed. Using the proposed method, the normal transformation for random variables with unknown cumulative distribution functions can be realized without using the Rosenblatt transformation. Through the numerical examples presented, the proposed method is found to be sufficiently accurate to include the random variables with unknown cumulative distribution functions in the first- and second-order reliability analyses with little extra computational effort.

Optical Observation of the Supercavitation Induced by High-Speed Water Entry

Abstract: When a high-speed projectile penetrates into water, a cavity is formed behind the projectile. The gas enclosed in the cavity experiences a nonequilibrium process, i.e., the gas pressure decreases as the projectile moves more deeply into water. As a result, the cavity is sealed near the free surface (surface closure) and subsequently the cavity breaks up in water (deep closure). Accompanying the break-up of the cavity, secondary shock waves appear. This is the so-called supercavitation in water entry. We describe an experimental investigation into the water entry phenomenon. Projectiles of 342 m/s were generated from a small-bore rifle that was fixed vertically in the experimental facility. The projectiles were fired into a windowed water tank. A shadowgraph optical observation was performed to observe the entry process of the projectile and the formation and collapse of the cavity behind the projectile. A number of interesting observations relating to the motion of the free surface, the splash, the underwater bubbly flow and so on were found

Randomized Leader Election Protocols in Radio Networks with No Collision Detection

Abstract: The main contribution of this work is to propose energy-efficient randomized leader election protocols for single-hop, single-channel radio networks (RN) that do not have collision detection (CD) capabilities We first presents a leader election protocol for the case the number n of stations is known beforehand The protocol runs in O(log f) time slots with no station being awake for more than O(log log f + log f/log n) time slots with probability at least 1 - 1/f for any f ≥ 1 We then present three leader election protocols for the case where n is not known beforehand The first protocol terminates, with probability exceeding 1 - 1/f, in O((log n)2 + (log f)2) time slots, with no station being awake for more than O(log n + log f) time slots Clearly, this first protocol terminates in O((log n)2) expected time slots Our second protocol reduces the expected termination time to O(log n) time slots This second protocol terminates, with probability exceeding 1 - 1/f in O(f3/5 log n) time slots Finally, by combining these two protocols, we obtain a third leader election protocol that terminates in O(log n) expected time slots This latter protocol terminates, with with probability exceeding 1 - 1/f, in O(min((log n)2 + (log f)2, f3/5log n)) time slots, with no station being awake for more than O(log n + log f) time slots

Energy-efficient initialization protocols for single-hop radio networks with no collision detection

Abstract: A radio network (RN, for short) is a distributed system consisting of n radio stations. We assume that the stations are small, bulk-produced, hand-held devices running on batteries and cannot be distinguished by serial or manufacturing number. Since recharging batteries may not be possible while on mission, we are interested in designing protocols that are highly energy-efficient. The initialization problem is to assign each of the n stations in the RN a unique ID. The initialization problem is nontrivial since the stations are assumed to be indistinguishable. The problem is fundamental, since practically all communication protocols for RNs proceed under the assumption that the RN has been initialized in advance. The main contribution of this work is to propose energy-efficient randomized initialization protocols for single-hop RNs lacking collision detection capabilities. First, we show that if the number n of stations is known beforehand, the single-channel RN can be initialized by a protocol that terminates, with probability exceeding 1-/sup 1///sub n/ in O(n) time slots, with no station being awake for more than O(log log n) time slots. We then go on to address the multichannel case and show that if k, (k/spl ges/1), channels are available, an n-station RN can be initialized, with probability exceeding 1-/sup 1///sub n/, in O(/sup n///sub k/+log n) time slots, with no station being awake for more than O(log log n) time slots.

Estimation of surface excitation correction factor for 200-5000 eV in Ni from absolute elastic scattering electron spectroscopy

Abstract: We have determined the surface plasmon excitation correction (SEC) factor for nickel in the 200-5000 eV range from the ratios of the absolute elastic scattering electron intensities measured by a novel cylindrical mirror analyser and those by the Monte Carlo method The inelastic mean free paths (IMFPs) of nickel used for the Monte Carlo method in the energy range specified were calculated by the Penn algorithm The resulting SECs were smaller than the values calculated from Chen and Oswald general equations of surface excitation parameters (SEPs), which describe the influence of surface plasmon excitations by electrons crossing a solid surface We also found that SEPs (obtained from SECs) could be fitted to the equation P s (α, E) = C/[E n cos(α) + C] or P s (α, E) = aE -b / cos(α) (<7% root-mean-square error) in the 200-5000 eV energy range, where P s is the SEP, a is the surface crossing angle of the electron to the surface normal, n(= 041), C(= 539), a(= 17) and b(= 029) are parameters and E is the electron energy

A comparative survey of synchronization algorithms for continuous media in network environments

Abstract: This paper makes a survey of algorithms proposed for continuous media synchronization in network environments. We classify media synchronization control techniques used in the algorithms into four categories: basic control, preventive control, reactive control and common control. We also pick four representatives from among the media synchronization algorithms and give outlines of them in order to explain how the techniques are employed in each algorithm. Furthermore, we make a comparison among 38 algorithms in terms of factors such as clocks, advance information on network delay bounds and synchronization control techniques, which determine the design of each algorithm.