Improving LVCSR system combination using neural network language model cross adaptation

State-of-the-art large vocabulary continuous speech recognition (LVCSR) systems often combine outputs from multiple subsystems developed at different sites. Cross system adaptation can be used as an alternative to direct hypothesis level combination schemes such as ROVER. The standard approach involves only cross adapting acoustic models. To fully exploit the complimentary features among sub-systems, language model (LM) cross adaptation techniques can be used. Previous research on multi-level n-gram LM cross adaptation is extended to further include the cross adaptation of neural network LMs in this paper. Using this improved LM cross adaptation framework, significant error rate gains of 4.0%-7.1% relative were obtained over acoustic model only cross adaptation when combining a range of Chinese LVCSR sub-systems used in the 2010 and 2011 DARPA GALE evaluations. Copyright © 2011 ISCA.

Coherent design methodology using modelling, simulation and optimisation

There is an increasing demand for optimising complete systems and the devices within that system, including capturing the interactions between the various multi-disciplinary (MD) components involved. Furthermore confidence in robust solutions is esential. As a consequence the computational cost rapidly increases and in many cases becomes infeasible to perform such conceptual designs. A coherent design methodology is proposed, where the aim is to improve the design process by effectively exploiting the potential of computational synthesis, search and optimisation and conventional simulation, with a reduction of the computational cost. This optimization framework consists of a hybrid optimization algorithm to handles multi-fidelity simulations. Simultaneously and in order to handles uncertainty without recasting the model and at affordable computational cost, a stochastic modelling method known as non-intrusive polynomial chaos is introduced. The effectiveness of the design methodology is demonstrated with the optimisation of a submarine propulsion system.

A Novel Passive UHF RFID Real Time Location Sensing System

Ubiquitous in-building Real Time Location Systems (RTLS) today are limited by costly active radio frequency identification (RFID) tags and short range portal readers of low cost passive RFID tags. We, however, present a novel technology locates RFID tags using a new approach based on (a) minimising RFID fading using antenna diversity, frequency dithering, phase dithering and narrow beam-width antennas, (b) measuring a combination of RSSI and phase shift in the coherent received tag backscatter signals and (c) being selective of use of information from the system by, applying weighting techniques to minimise error. These techniques make it possible to locate tags to an accuracy of less than one metre. This breakthrough will enable, for the first time, the low-cost tagging of items and the possibility of locating them at relatively high precision.

Control of a shock-wave/boundary-layer interaction and subsequent subsonic diffuser using a combination of vortex generators and bleed

An experimental investigation has been undertaken in which vortex generators (VGs) have been employed to inhibit boundary-layer separation produced by the combined adversepressure- gradient of a terminal shock-wave and subsonic diffuser. This setup has been developed as part of a program to produce a more inlet relevant flow-field using a small-scale wind tunnel than previous studies. The resulting flow is dominated by large-scale separation, and as such, is thought to be a good test-bed for flow control. In this investigation, VGs have been added to determine their potential for shock-induced separation mitigation. In line with previous studies, it was observed that the application of VGs alone was not able to significantly alleviate separation overall, because enlarged corner separations was observed. Only when control of the corner separations using corner bleed was employed alongside centre-span control using VGs was a significant improvement in both wall pressure recovery (6% increase) and stagnation pressure recovery (2.4% increase) observed. Copyright © 2012 by the American Institute of Aeronautics and Astronautics, Inc.

Language model cross adaptation for LVCSR system combination

State-of-the-art large vocabulary continuous speech recognition (LVCSR) systems often combine outputs from multiple sub-systems that may even be developed at different sites. Cross system adaptation, in which model adaptation is performed using the outputs from another sub-system, can be used as an alternative to hypothesis level combination schemes such as ROVER. Normally cross adaptation is only performed on the acoustic models. However, there are many other levels in LVCSR systems' modelling hierarchy where complimentary features may be exploited, for example, the sub-word and the word level, to further improve cross adaptation based system combination. It is thus interesting to also cross adapt language models (LMs) to capture these additional useful features. In this paper cross adaptation is applied to three forms of language models, a multi-level LM that models both syllable and word sequences, a word level neural network LM, and the linear combination of the two. Significant error rate reductions of 4.0-7.1% relative were obtained over ROVER and acoustic model only cross adaptation when combining a range of Chinese LVCSR sub-systems used in the 2010 and 2011 DARPA GALE evaluations. © 2012 Elsevier Ltd. All rights reserved.

Shock wave/boundary-layer interaction control using a combination of vortex generators and bleed

To investigate whether vortex generators can be an effective form of passive flow control an experimental investigation has been conducted in a small-scale wind tunnel. With specific emphasis on supersonic inlet applications flow separation was initiated using a combined terminal shock wave and subsonic diffuser: a configuration that has been developed as a part of a program to produce a more inlet-relevant flowfield in a small-scale wind tunnel than previous studies. When flow control was initially introduced little overall flow improvement was obtained as the losses tended to be redistributed instead of removed. It became apparent that there existed a strong coupling between the center-span flow and the corner flows. As a consequence, only when flow control was applied to both the corner flows and center-span flow was a significant flow improvement obtained. When corner suction and center-span vortex generators were employed in tandem separation was much reduced and wall-pressure and stagnation pressure were notably improved. As a result, when applied appropriately, it is thought that vortex generators do have the potential to reduce the dependence on boundary-layer bleed for the purpose of separation suppression. Copyright © 2012 by Neil Titchener and Holger Babinsky. Published by the American Institute of Aeronautics and Astronautics, Inc.

Design of diesel smoke feedback control using a combination of PI control algorithm and performance optimization

A novel smoke sensor was used to realize smoke feedback control on a diesel engine. The controller design based on a combination of PI control algorithm and the engine performance optimization is described. Experimental results demonstrate how this control system behave to meet both of the speed and smoke requirements during engine transients.

Role of a high gain of the medium in superradiance generation and in observation of coherent effects in semiconductor lasers

A theoretical model of superradiant pulse generation in semiconductor laser structures is developed. It is shown that a high optical gain of the medium can overcome phase relaxation and results in a built-up superradiant state (macroscopic dipole) in an assembly of electron - hole pairs on a time scale much longer than the characteristic polarisation relaxation time T2. A criterion of the superradiance generation is the condition acmT2 > 1, where α is the gain coefficient and cm is the speed of light in the medium. The theoretical model describes both qualitatively and quantitatively the author's own experimental results.

System combination and score normalization for spoken term detection

Spoken content in languages of emerging importance needs to be searchable to provide access to the underlying information. In this paper, we investigate the problem of extending data fusion methodologies from Information Retrieval for Spoken Term Detection on low-resource languages in the framework of the IARPA Babel program. We describe a number of alternative methods improving keyword search performance. We apply these methods to Cantonese, a language that presents some new issues in terms of reduced resources and shorter query lengths. First, we show score normalization methodology that improves in average by 20% keyword search performance. Second, we show that properly combining the outputs of diverse ASR systems performs 14% better than the best normalized ASR system. © 2013 IEEE.