Robust noise reduction for speech and audio signals

Statistical model-based methods are presented for the reconstruction of autocorrelated signals in impulsive plus continuous noise environments. Signals are modelled as autoregressive and noise sources as discrete and continuous mixtures of Gaussians, allowing for robustness in highly impulsive and non-Gaussian environments. Markov Chain Monte Carlo methods are used for reconstruction of the corrupted waveforms within a Bayesian probabilistic framework and results are presented for contaminated voice and audio signals.

A study of matrix analysis methods for inelastic structures. Project no. 1467, Task no. 146701, prepared under Contract no. AF 33 (657)-8717 by Massachusetts Institute of Technology, Aeroelastic and Structures Research Laboratory.

Cover title.

QUT speaker identity verification system for EVALITA 2009

This document outlines the system submitted by the Speech and Audio Research Laboratory at the Queensland University of Technology (QUT) for the Speaker Identity Verication: Application task of EVALITA 2009. This submission consisted of a score-level fusion of three component systems, a joint-factor GMM system and two SVM systems using GLDS and GMM supervector kernels. Development and evaluation results are presented, demonstrating the effectiveness of this fused system approach.

QUT Speaker Identity Verification system for EVALITA 2009

This document outlines the system submitted by the Speech and Audio Research Laboratory at the Queensland University of Technology (QUT) for the Speaker Identity Verification: Application task of EVALITA 2009. This competitive submission consisted of a score-level fusion of three component systems; a joint-factor analysis GMM system and two SVM systems using GLDS and GMM supervector kernels. Development evaluation and post-submission results are presented in this study, demonstrating the effectiveness of this fused system approach. This study highlights the challenges associated with system calibration from limited development data and that mismatch between training and testing conditions continues to be a major source of error in speaker verification technology.

Enhancing teaching and learning speech acts: action research of Japanese as a foreign language /

This action research observes a second year Japanese class at a university where foreign language courses are elective for undergraduate students. In this study, using the six strategies to teach Japanese speech acts that Ishihara and Cohen (2006) suggested, I conducted three classes and analyzed my teaching practice with a critical friend. These strategies assist learners toward the development of their understanding of the following Japanese speech acts and also keep the learners to use them in a manner appropriate to the context: (I) invitation and refusal; (2) compliments; and (3) asking for a permission. The aim of this research is not only to improve my instruction in relation to second language (L2) pragmatic development, but also to raise further questions and to develop future research. The findings are analyzed and the data derived from my journals, artifacts, students' work, observation sheets, interviews with my critical friend, and pretests and posttests are coded and presented. The analysis shows that (I) after my critical friend encouraged my study and my students gave me some positive comments after each lesson, I gained confidence in teaching the suggested speech acts; (2) teaching involved explaining concepts and strategies, creating the visual material (a video) showing the strategies, and explaining the relationship between the strategy and grammatical forms and samples of misusing the forms; (3) students' background and learning styles influenced lessons; and (4) pretest and posttests showed that the students' Icvel of their L2 appropriate pragmatics dramatically improved after each instruction. However, after careful observation, it was noted that some factors prevented students from producing the correct output even though they understood the speech act differences.

Viral infections in workers in hospital and research laboratory settings: a comparative review of infection modes and respective biosafety aspects

Objectives: To compare modes and sources of infection and clinical and biosafety aspects of accidental viral infections in hospital workers and research laboratory staff reported in scientific articles. Methods: PubMed, Google Scholar, ISI Web of Knowledge, Scirus, and Scielo were searched (to December 2008) for reports of accidental viral infections, written in English, Portuguese, Spanish, or German; the authors' personal file of scientific articles and references from the articles retrieved in the initial search were also used. Systematic review was carried out with inclusion criteria of presence of accidental viral infection's cases information, and exclusion criteria of absence of information about the viral etiology, and at least probable mode of infection.Results: One hundred and forty-one scientific articles were obtained, 66 of which were included in the analysis. For arboviruses, 84% of the laboratory infections had aerosol as the source; for alphaviruses alone, aerosol exposure accounted for 94% of accidental infections. of laboratory arboviral infections, 15.7% were acquired percutaneously, whereas 41.6% of hospital infections were percutaneous. For airborne viruses, 81% of the infections occurred in laboratories, with hantavirus the leading causative agent. Aerosol inhalation was implicated in 96% of lymphocytic choriomeningitis virus infections, 99% of hantavirus infections, and 50% of coxsackievirus infections, but infective droplet inhalation was the leading mode of infection for severe acute respiratory syndrome coronavirus and the mucocutaneous mode of infection was involved in the case of infection with influenza B. For blood-borne viruses, 92% of infections occurred in hospitals and 93% of these had percutaneous mode of infection, while among laboratory infections 77% were due to infective aerosol inhalation. Among blood-borne virus infections there were six cases of particular note: three cases of acute hepatitis following hepatitis C virus infection with a short period of incubation, one laboratory case of human immunodeficiency virus infection through aerosol inhalation, one case of hepatitis following hepatitis G virus infection, and one case of fulminant hepatitis with hepatitis B virus infection following exposure of the worker's conjunctiva to hepatitis B virus e antigen-negative patient saliva. of the 12 infections with viruses with preferential mucocutaneous transmission, seven occurred percutaneously, aerosol was implicated as a possible source of infection in two cases, and one atypical infection with Macacine herpesvirus 1 with fatal encephalitis as the outcome occurred through a louse bite. One outbreak of norovirus infection among hospital staff had as its probable mode of infection the ingestion of inocula spread in the environment by fomites.Conclusions: The currently accepted and practiced risk analysis of accidental viral infections based on the conventional dynamics of infection of the etiological agents is insufficient to cope with accidental viral infections in laboratories and to a lesser extent in hospitals, where unconventional modes of infection are less frequently present but still have relevant clinical and potential epidemiological consequences. Unconventional modes of infection, atypical clinical development, or extremely severe cases are frequently present together with high viral loads and high virulence of the agents manipulated in laboratories. In hospitals by contrast, the only possible association of atypical cases is with the individual resistance of the worker. Current standard precaution practices are insufficient to prevent most of the unconventional infections in hospitals analyzed in this study; it is recommended that special attention be given to flaviviruses in these settings. (C) 2011 International Society for Infectious Diseases. Published by Elsevier Ltd. All rights reserved.