Factor analysis based VTS and JUD noise estimation and compensation

Model based compensation schemes are a powerful approach for noise robust speech recognition. Recently there have been a number of investigations into adaptive training, and estimating the noise models used for model adaptation. This paper examines the use of EM-based schemes for both canonical models and noise estimation, including discriminative adaptive training. One issue that arises when estimating the noise model is a mismatch between the noise estimation approximation and final model compensation scheme. This paper proposes FA-style compensation where this mismatch is eliminated, though at the expense of a sensitivity to the initial noise estimates. EM-based discriminative adaptive training is evaluated on in-car and Aurora4 tasks. FA-style compensation is then evaluated in an incremental mode on the in-car task. © 2011 IEEE.

Rapid joint speaker and noise compensation for robust speech recognition

For speech recognition, mismatches between training and testing for speaker and noise are normally handled separately. The work presented in this paper aims at jointly applying speaker adaptation and model-based noise compensation by embedding speaker adaptation as part of the noise mismatch function. The proposed method gives a faster and more optimum adaptation compared to compensating for these two factors separately. It is also more consistent with respect to the basic assumptions of speaker and noise adaptation. Experimental results show significant and consistent gains from the proposed method. © 2011 IEEE.

Model-based approaches to handling additive noise in reverberant environments

Model-based approaches to handle additive and convolutional noise have been extensively investigated and used. However, the application of these schemes to handling reverberant noise has received less attention. This paper examines the extension of two standard additive/convolutional noise approaches to handling reverberant noise. The first is an extension of vector Taylor series (VTS) compensation, reverberant VTS, where a mismatch function including reverberant noise is used. The second scheme modifies constrained MLLR to allow a wide-span of frames to be taken into account and projected into the required dimensionality. To allow additive noise to be handled, both these schemes are combined with standard VTS. The approaches are evaluated and compared on two tasks, MC-WSJ-AV, and a reverberant simulated version of AURORA-4. © 2011 IEEE.

Synchronization in a pair of thermally coupled rotating baroclinic annuli: understanding atmospheric teleconnections in the laboratory.

Synchronization phenomena in a fluid dynamical analogue of atmospheric circulation is studied experimentally by investigating the dynamics of a pair of thermally coupled, rotating baroclinic annulus systems. The coupling between the systems is in the well-known master-slave configuration in both periodic and chaotic regimes. Synchronization tools such as phase dynamics analysis are used to study the dynamics of the coupled system and demonstrate phase synchronization and imperfect phase synchronization, depending upon the coupling strength and parameter mismatch.

On-line policy optimisation of spoken dialogue systems via live interaction with human subjects

Statistical dialogue models have required a large number of dialogues to optimise the dialogue policy, relying on the use of a simulated user. This results in a mismatch between training and live conditions, and significant development costs for the simulator thereby mitigating many of the claimed benefits of such models. Recent work on Gaussian process reinforcement learning, has shown that learning can be substantially accelerated. This paper reports on an experiment to learn a policy for a real-world task directly from human interaction using rewards provided by users. It shows that a usable policy can be learnt in just a few hundred dialogues without needing a user simulator and, using a learning strategy that reduces the risk of taking bad actions. The paper also investigates adaptation behaviour when the system continues learning for several thousand dialogues and highlights the need for robustness to noisy rewards. © 2011 IEEE.

Mismatched shaping schemes for bit-interleaved coded modulation

We consider bit-interleaved coded modulation (BICM) schemes where, instead of the true bit or symbol probabilities and the constellation used at the transmitter, the decoder uses arbitrary probabilities or reference constellations. We study the corresponding low- and high- signal-to-noise-ratio regimes and show that even in the presence of this extra sources of mismatch, BICM has a negligible penalty with respect to coded modulation. © 2012 IEEE.

Failure mechanisms in brittle laminates

Steady-state tunneling and plane-strain delamination of an H-shape crack are examined for elastic, isotropic multi layers. Both tunneling and delamination are analysed by employing linear elastic fracture mechanics within a 2D finite element framework. Failure maps are produced to reveal the sensitivity of cracking path to the relative toughness of layer and interface, and to the stiffness mismatch of layers.

Correlations between the morphology and emission properties of trench defects in InGaN/GaN quantum wells

Atomic force microscopy (AFM) and scanning electron microscopy (SEM) with cathodoluminescence (CL) were performed on exactly the same defects in a blue-emitting InGaN/GaN multiple quantum well (QW) sample enabling the direct correlation of the morphology of an individual defect with its emission properties. The defects in question are observed in AFM and SEM as a trench partially or fully enclosing a region of the QW having altered emission properties. Their sub-surface structure has previously been shown to consist of a basal plane stacking fault (BSF) in the plane of the QW stack, and a stacking mismatch boundary (SMB) which opens up into a trench at the sample surface. In CL, the material enclosed by the trench may emit more or less intensely than the surrounding material, but always exhibits a redshift relative to the surrounding material. A strong correlation exists between the width of the trench and both the redshift and the intensity ratio, with the widest trenches surrounding regions which exhibit the brightest and most redshifted emission. Based on studies of the evolution of the trench width with the number of QWs from four additional MQW samples, we conclude that in order for a trench defect to emit intense, strongly redshifted light, the BSF must be formed in the early stages of the growth of the QW stack. The data suggest that the SMB may act as a non-radiative recombination center. © 2013 American Institute of Physics.

Understanding the kink formation in GaAs/InAs heterostructural nanowires

The kinks formation in heterostructural nanowires was observed to be dominant when InAs nanowires were grown on GaAs nanowires. Nanowires were grown through vapor-liquid-solid (VLS) mechanism in an MOCVD (metalorganic chemical vapor deposition) reactor. GaAs nanowires were grown in [1 1 1 ]B direction on a GaAs (1 1 1 )B substrate. When InAs nanowires grown on the GaAs nanowires, most of the InAs nanowires changed their growth directions from [1 1 1 ]B to other 〈111〉B directions. The kinks formation is ascribed to the large compressive misfit strain at the GaAs/InAs interface (7.2% lattice mismatch between GaAs and InAs) and the high mobility of indium species during MOCVD growth. The in-depth analysis of the kinks formation was done by growing InAs for short times on the GaAs nanowires and characterizing the samples. The hindrance to compressively strain InAs to form coherent layers with GaAs pushed the InAs/Au interfaces to the sides of the GaAs nanowires growth ends. New InAs/Au interfaces have generated at the sides of GaAs nanowires, due to lateral growth of InAs on GaAs nanowires. These new interfaces led the InAs nanowires growth in other 〈111〉B directions. The morphological and structural features of these heterostructural kinked nanowires were characterized using scanning electron microscopy (SEM) and transmission electron microscopy (TEM) techniques. © 2006 IEEE.

Temperature dependence of photoluminescence from single core-shell GaAs-AlGaAs nanowires

Temperature-dependent polarized microphotoluminescence measurements of single GaAsAlGaAs core-shell nanowires are used to probe their electronic states. The low-temperature emission from these wires is strongly enhanced compared with that observed in bare GaAs nanowires and is strongly polarized, reflecting the dielectric mismatch between the nanowire and the surrounding air. The temperature-dependent band gap of the nanowires is seen to be somewhat different from that observed in bulk GaAs, and the PL rapidly quenches above 120 K, with an activation energy of 17 meV reflecting the presence of nonradiative defects. © 2006 American Institute of Physics.

Growth of GaAs/InAs vertical nanowires on GaAs (111)B by metalorganic chemical vapor deposition

The growth mechanism and properties of GaAs/InAs nanowires prepared by metalorganic chemical vapor deposition are investigated. Vertical InAs nanowires on GaAs (111)B substrates are successfully grown despite the large lattice mismatch (-7.2%). The crystallographic perfection of InAs nanowires is confirmed by hexagonal or triangular cross section. An interesting L-shaping of GaAs/InAs heterostructure nanowire which could be useful for novel device application is observed. © 2005 IEEE.

Investigation of a radial-inflow bleed as a potential for compressor clearance control

The mismatch in thermal response between a High Pressure Compressor (HPC) drum and casing is a limiting factor in the reduction of compressor clearance. An experimental test rig has been used to demonstrate the concept of radial inflow to reduce the thermal time constant of HPC discs. The testing uses a simulated idle - Maximum Take Off (MTO) - idle transient in order to measure the thermal response directly. The testing is fully scaled in the dimensionless sense to engine conditions. A simple closure model based on lumped capacitance is used to illustrate the scope of potential benefits. The proof-of-concept testing shows that HPC disc time constant reductions of the order 2 are feasible with a radial-inflow bleed of only 4% of bore flow at scaled MTO conditions. Using the experimental results, the simple closure modelling suggests that for a stage with a significant mismatch in thermal response, reductions in 2D axis-symmetric clearance of as much as 50% at MTO conditions may be possible along with significant scope for improvements at cruise conditions. Copyright © 2013 by ASME.