Compression techniques applied to multiple speech recognition systems

Speech recognition systems typically contain many Gaussian distributions, and hence a large number of parameters. This makes them both slow to decode speech, and large to store. Techniques have been proposed to decrease the number of parameters. One approach is to share parameters between multiple Gaussians, thus reducing the total number of parameters and allowing for shared likelihood calculation. Gaussian tying and subspace clustering are two related techniques which take this approach to system compression. These techniques can decrease the number of parameters with no noticeable drop in performance for single systems. However, multiple acoustic models are often used in real speech recognition systems. This paper considers the application of Gaussian tying and subspace compression to multiple systems. Results show that two speech recognition systems can be modelled using the same number of Gaussians as just one system, with little effect on individual system performance. Copyright © 2009 ISCA.

Oscillatory sensitivity patterns for global modes in wakes

Globally unstable wakes with co-flow at intermediate Reynolds numbers are studied, to quantify important spatial regions for the development and control of the global instability. One region of high structural sensitivity is found close to the inlet for all wakes, in agreement with previous findings for cylinder wakes. A second, elongated region of high structural sensitivity is seen downstream of the first one for unconfined wakes at Re = 400. When base flow modifications are considered, a spatially oscillating sensitivity pattern is found inside the downstream high structural sensitivity region. This implies that the same change in the base flow can either destabilize or stabilize the flow, depending on the exact position where it is applied. It is shown that the sensitivity pattern remains unchanged for different choices of streamwise boundary conditions and numerical resolution. Actual base flow modifications are applied in selected configurations, and the linear global modes recomputed. It is confirmed that the linear global eigenvalues move according to the predicted sensitivity pattern for small amplitude base flow modifications, for which the theory applies. We also look at the implications of a small control cylinder on the flow. Only the upstream high sensitivity region proves to be robust in terms of control, but one should be careful not to disturb the flow in the downstream high sensitivity region, in order to achieve control. The findings can have direct implications on the numerical resolution requirements for wakes at higher Reynolds numbers. Furthermore, they provide one more possible explanation to why confined wakes have a more narrow frequency spectrum than unconfined wakes.

A probabilistic energy model for non-domestic building sectors applied to analysis of school buildings in greater London

The diversity of non-domestic buildings at urban scale poses a number of difficulties to develop models for large scale analysis of the stock. This research proposes a probabilistic, engineering-based, bottom-up model to address these issues. In a recent study we classified London's non-domestic buildings based on the service they provide, such as offices, retail premise, and schools, and proposed the creation of one probabilistic representational model per building type. This paper investigates techniques for the development of such models. The representational model is a statistical surrogate of a dynamic energy simulation (ES) model. We first identify the main parameters affecting energy consumption in a particular building sector/type by using sampling-based global sensitivity analysis methods, and then generate statistical surrogate models of the dynamic ES model within the dominant model parameters. Given a sample of actual energy consumption for that sector, we use the surrogate model to infer the distribution of model parameters by inverse analysis. The inferred distributions of input parameters are able to quantify the relative benefits of alternative energy saving measures on an entire building sector with requisite quantification of uncertainties. Secondary school buildings are used for illustrating the application of this probabilistic method. © 2012 Elsevier B.V. All rights reserved.

A subdivision-based implementation of the hierarchical b-spline finite element method

A novel technique is presented to facilitate the implementation of hierarchical b-splines and their interfacing with conventional finite element implementations. The discrete interpretation of the two-scale relation, as common in subdivision schemes, is used to establish algebraic relations between the basis functions and their coefficients on different levels of the hierarchical b-spline basis. The subdivision projection technique introduced allows us first to compute all element matrices and vectors using a fixed number of same-level basis functions. Their subsequent multiplication with subdivision matrices projects them, during the assembly stage, to the correct levels of the hierarchical b-spline basis. The proposed technique is applied to convergence studies of linear and geometrically nonlinear problems in one, two and three space dimensions. © 2012 Elsevier B.V.

Deformation and fracture of impulsively loaded sandwich panels

Light metal sandwich panel structures with cellular cores have attracted interest for multifunctional applications which exploit their high bend strength and impact energy absorption. This concept has been explored here using a model 6061-T6 aluminum alloy system fabricated by friction stir weld joining extruded sandwich panels with a triangular corrugated core. Micro-hardness and miniature tensile coupon testing revealed that friction stir welding reduced the strength and ductility in the welds and a narrow heat affected zone on either side of the weld by approximately 30%. Square, edge clamped sandwich panels and solid plates of equal mass per unit area were subjected to localized impulsive loading by the impact of explosively accelerated, water saturated, sand shells. The hydrodynamic load and impulse applied by the sand were gradually increased by reducing the stand-off distance between the test charge and panel surfaces. The sandwich panels suffered global bending and stretching, and localized core crushing. As the pressure applied by the sand increased, face sheet fracture by a combination of tensile stretching and shear-off occurred first at the two clamped edges of the panels that were parallel with the corrugation and weld direction. The plane of these fractures always lay within the heat affected zone of the longitudinal welds. For the most intensively loaded panels additional cracks occurred at the other clamped boundaries and in the center of the panel. To investigate the dynamic deformation and fracture processes, a particle-based method has been used to simulate the impulsive loading of the panels. This has been combined with a finite element analysis utilizing a modified Johnson-Cook constitutive relation and a Cockcroft-Latham fracture criterion that accounted for local variation in material properties. The fully coupled simulation approach enabled the relationships between the soil-explosive test charge design, panel geometry, spatially varying material properties and the panel's deformation and dynamic failure responses to be explored. This comprehensive study reveals the existence of a strong instability in the loading that results from changes in sand particle reflection during dynamic evolution of the panel's surface topology. Significant fluid-structure interaction effects are also discovered at the sample sides and corners due to changes of the sand reflection angle by the edge clamping system. © 2012 Elsevier Ltd. All rights reserved.

The fluid mechanics of a two-dimensional Aaberg exhaust hood

In this paper a study of the air flow pattern created by a two-dimensional Aaberg exhaust hood local ventilation system is presented. A mathematical model of the flow, in terms of the stream function ψ, is derived analytically for both laminar and turbulent injections of fluid. Streamlines and lines of constant speed deduced from the model are examined for various values of the governing dimensionless operating parameter and predictions are given as to the area in front of the hood from which the air can be sampled. The effect of the injection of fluid on the centre-line velocity of the flow is examined and a comparison of the results with the available experimental data is given. © 1992.