Speech recognition in living rooms: Integrated speech enhancement and recognition system based on spatial,spectral and temporal modeling of sounds

Research on noise robust speech recognition has mainly focused on dealing with relatively stationary noise that may differ from the noise conditions in most living environments. In this paper, we introduce a recognition system that can recognize speech in the presence of multiple rapidly time-varying noise sources as found in a typical family living room. To deal with such severe noise conditions, our recognition system exploits all available information about speech and noise; that is spatial (directional), spectral and temporal information. This is realized with a model-based speech enhancement pre-processor, which consists of two complementary elements, a multi-channel speech–noise separation method that exploits spatial and spectral information, followed by a single channel enhancement algorithm that uses the long-term temporal characteristics of speech obtained from clean speech examples. Moreover, to compensate for any mismatch that may remain between the enhanced speech and the acoustic model, our system employs an adaptation technique that combines conventional maximum likelihood linear regression with the dynamic adaptive compensation of the variance of the Gaussians of the acoustic model. Our proposed system approaches human performance levels by greatly improving the audible quality of speech and substantially improving the keyword recognition accuracy.     

A Single Mutation at the Ferredoxin Binding Site of P450 Vdh Enables Efficient Biocatalytic Production of 25‐Hydroxyvitamin D3

Vitamin D₃ hydroxylase (Vdh) from Pseudonocardia autotrophica is a cytochrome P450 monooxygenase that catalyzes the two‐step hydroxylation of vitamin D₃ (VD₃) to produce 25‐hydroxyvitamin D₃ (25(OH)VD₃) and 1α,25‐dihydroxyvitamin D₃ (1α,25(OH)₂VD₃). These hydroxylated forms of VD₃ are useful as pharmaceuticals for the treatment of conditions associated with VD₃ deficiency and VD₃ metabolic disorder. Herein, we describe the creation of a highly active T107A mutant of Vdh by engineering the putative ferredoxin‐binding site. Crystallographic and kinetic analyses indicate that the T107A mutation results in conformational change from an open to a closed state, thereby increasing the binding affinity with ferredoxin. We also report the efficient biocatalytic synthesis of 25(OH)VD₃, a promising intermediate for the synthesis of various hydroxylated VD₃ derivatives, by using nisin‐treated Rhodococcus erythropolis cells containing Vdh_T107A. The gene‐expression cassette encoding Bacillus megaterium glucose dehydrogenase‐IV was inserted into the R. erythropolis chromosome and expressed to avoid exhaustion of NADH in a cytoplasm during bioconversion. As a result, approximately 573 μg mL^?1 25(OH)VD₃ was successfully produced by a 2 h bioconversion.     

Fabrication of vertically aligned single-walled carbon nanotubes in atmospheric pressure non-thermal plasma CVD

A fabrication technique of high-purity vertically aligned single-walled carbon nanotubes (VA-SWCNTs) using atmospheric pressure plasma enhanced chemical vapor deposition is presented. Although densely mono-dispersed Fe-Co catalysts of a few nanometers is primarily responsible for VA-SWCNT growth, carbon precipitation was virtually absent in the thermal CVD regime at 700 °C. On the other hand, high-purity VA-SWCNTs without measurable defects were grown at 4 μm min⁻¹ by applying atmospheric pressure radio-frequency discharge (APRFD) which has been previously developed for this purpose. The results proved that cathodic ion sheath adjacent to the substrates, where a large potential drop exists, also plays an essential role for the controlled growth of SWCNTs, while ion damage to the VA-SWCNTs is inherently avoided due to high collision frequency among molecules in atmospheric pressure. Operation regime of APRFD and tentative reaction mechanisms for VA-SWCNT growth are discussed along with optical emission spectroscopy of near substrate region.     

An energy trading system with consideration of CO2 emissions

Abatement of CO₂ emission is one of the most important issues in the 21st century regarding preservation of the earth environment. This paper addresses a utility operations planning problem for distributed energy management systems (DEMSs), where we are to obtain optimal plans that minimize both costs and CO₂ emissions. A DEMS consists of multiple entities that seek their own economic profits. In this paper, we give a mathematical formulation of the utility operations planning problem for each entity, and propose an energy trading market, which utlizes a multi‐attribute auction protocol in order to deal with both a price and a CO₂ emission rate. Experimental results show that collaboration among entities through the market provides a more profitable plan for each entity and abatement of CO₂ emission is also achieved. © 2007 Wiley Periodicals, Inc. Electr Eng Jpn, 162(4): 54–63, 2008; Published online in Wiley InterScience ( www.interscience.wiley. com ). DOI 10.1002/eej.20418     

Forecasting electric power generation in a photovoltaic power system for an energy network

Recently, there has been an increase in concern about the global environment. Interest is growing in developing an energy network by which new energy systems such as photovoltaic and fuel cells generate power locally and electrical power and heat are controlled with a communication network. We developed the power generation forecast method for photovoltaic power systems in an energy network. The method makes use of weather information and regression analysis. We carried out forecasting power output of the photovoltaic power system installed in Expo 2005, Aichi Japan. As a result of comparing measurements with prediction values, the average prediction error per day was about 26% of the measured power. © 2009 Wiley Periodicals, Inc. Electr Eng Jpn, 167(4): 16–23, 2009; Published online in Wiley InterScience ( www.interscience.wiley.com ). DOI 10.1002/eej.20755     

Numerical analysis of characteristics of microactuators driven by liquid crystals

For the purpose of developing liquid crystalline microactuators, the transient behaviors of a nematic liquid crystal between two parallel plates have been computed for various parameters such as applied voltage, the gap between the plates, and the twist and tilt angles at the plates. The Leslie-Ericksen theory has been selected as a constitutive equation. The twist angle has an effect on the induced velocity profiles; for example, the induced flow is planar at the twist angle of 0°, while the flow has an out-of-plane component when the twist angle is not 0°. Transient behaviors of shear stress acting on the plates, the flowrate, and the maximum values of the velocity, and the tilt angle between the plates have been reported. In addition, we have investigated the effects of the applied voltage, the gap between the plates, and the tilt angle at the plates on the above-mentioned values. We can develop microactuators with arbitrary characteristics by suitably controlling the applied voltage, the size of the actuators, and the director anchoring conditions.     

Improvement of anode oxidation reaction of a fuel cell using ammonium formate with Pt‐Ir catalysis

We have attempted to develop high‐performance and safe fuel cells by using ammonium formate as a solid (powder) fuel. This solid fuel has the potential of safer transportability than liquid fuels such as methanol from the viewpoint of toxicity and flammability. In order to make use of some of the advantages of ammonium formate, we investigated the oxidation characteristics of ammonium formate with respect to a Pt electrode. Cyclic voltammograms indicate that ammonium formate has high oxidation activity with respect to a Pt electrode. We have also found that the oxidation of ammonium formate can be improved by the addition of Ir to Pt catalysts. It is highly likely that ammonium formate will be useful as a solid (powder) fuel for polymer electrolyte membrane (PEM) fuel cells. This new fuel will promote the development of safe fuel cells for PEM. © 2010 Wiley Periodicals, Inc. Electr Eng Jpn, 174(4): 45–50, 2011; Published online in Wiley Online Library ( wileyonlinelibrary.com ). DOI 10.1002/eej.21043     

Optimal memoryless regulator of large‐scale systems with time delays in the interconnections and the states via decentralized control

A method to construct optimal memoryless regulators of large‐scale systems with time delays in the interconnections and the states is proposed. Each feedback gain of decentralized control laws is obtained directly from solutions of simultaneous linear matrix inequalities. First, a sufficient condition for the stability of the overall closed‐ loop system is presented. Then it is extended so that the overall system and each subsystem remain stable, even if all or some of the interconnections between subsystems are cut. Lastly, a simplification of the implementation is discussed in the perspective of feedback loop reduction. © 2011 Wiley Periodicals, Inc. Electr Eng Jpn, 177(4): 42–50, 2011; Published online in Wiley Online Library ( wileyonlinelibrary.com ). DOI 10.1002/eej.21191     

Fabrication of submicron alumina ceramics by pulse electric current sintering using M2+ (M = Mg,Ca, Ni)-doped alumina nanopowders

Dense submicron-grained alumina ceramics were fabricated by pulse electric current sintering (PECS) using M²⁺(M: Mg, Ca, Ni)-doped alumina nanopowders at 1250 °C under a uniaxial pressure of 80 MPa. The M²⁺-doped alumina nanopowders (0–0.10 mass%) were prepared through a new sol–gel route using high-purity polyhydroxoaluminum (PHA) and MCl₂ solutions as starting materials. The composite gels obtained were calcined at 900 °C and ground by planetary ball milling. The powders were re-calcined at 900 °C to increase the content of α-alumina particles, which act as seeding for low-temperature densification. Densification and microstructural development depend on the M²⁺ dopant species. Dense alumina ceramics (relative density ≥99.0%) thus obtained had a uniform microstructure composed of fine grains, where the average grain size developed for non-doped, Ni-doped, Mg-doped and Ca-doped samples was 0.67, 0.67, 0.47 and 0.30 μm, respectively, showing that Ca-doping is the most promising method for tailoring of nanocrystalline alumina ceramics.