377 resultados para speaker identification


Relevância:

20.00% 20.00%

Publicador:

Resumo:

Speaker diarization is the process of annotating an input audio with information that attributes temporal regions of the audio signal to their respective sources, which may include both speech and non-speech events. For speech regions, the diarization system also specifies the locations of speaker boundaries and assign relative speaker labels to each homogeneous segment of speech. In short, speaker diarization systems effectively answer the question of ‘who spoke when’. There are several important applications for speaker diarization technology, such as facilitating speaker indexing systems to allow users to directly access the relevant segments of interest within a given audio, and assisting with other downstream processes such as summarizing and parsing. When combined with automatic speech recognition (ASR) systems, the metadata extracted from a speaker diarization system can provide complementary information for ASR transcripts including the location of speaker turns and relative speaker segment labels, making the transcripts more readable. Speaker diarization output can also be used to localize the instances of specific speakers to pool data for model adaptation, which in turn boosts transcription accuracies. Speaker diarization therefore plays an important role as a preliminary step in automatic transcription of audio data. The aim of this work is to improve the usefulness and practicality of speaker diarization technology, through the reduction of diarization error rates. In particular, this research is focused on the segmentation and clustering stages within a diarization system. Although particular emphasis is placed on the broadcast news audio domain and systems developed throughout this work are also trained and tested on broadcast news data, the techniques proposed in this dissertation are also applicable to other domains including telephone conversations and meetings audio. Three main research themes were pursued: heuristic rules for speaker segmentation, modelling uncertainty in speaker model estimates, and modelling uncertainty in eigenvoice speaker modelling. The use of heuristic approaches for the speaker segmentation task was first investigated, with emphasis placed on minimizing missed boundary detections. A set of heuristic rules was proposed, to govern the detection and heuristic selection of candidate speaker segment boundaries. A second pass, using the same heuristic algorithm with a smaller window, was also proposed with the aim of improving detection of boundaries around short speaker segments. Compared to single threshold based methods, the proposed heuristic approach was shown to provide improved segmentation performance, leading to a reduction in the overall diarization error rate. Methods to model the uncertainty in speaker model estimates were developed, to address the difficulties associated with making segmentation and clustering decisions with limited data in the speaker segments. The Bayes factor, derived specifically for multivariate Gaussian speaker modelling, was introduced to account for the uncertainty of the speaker model estimates. The use of the Bayes factor also enabled the incorporation of prior information regarding the audio to aid segmentation and clustering decisions. The idea of modelling uncertainty in speaker model estimates was also extended to the eigenvoice speaker modelling framework for the speaker clustering task. Building on the application of Bayesian approaches to the speaker diarization problem, the proposed approach takes into account the uncertainty associated with the explicit estimation of the speaker factors. The proposed decision criteria, based on Bayesian theory, was shown to generally outperform their non- Bayesian counterparts.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

This study examined primary school teachers’ knowledge of anxiety and excessive anxiety symptoms in children. Three hundred and fifteen primary school teachers completed a questionnaire exploring their definitions of anxiety and the indications they associated with excessive anxiety in primary school children. Results showed that teachers had an understanding of what anxiety was in general but did not consistently distinguish normal anxiety from excessive anxiety, often defining all anxiety as a negative experience. Teachers were able to identify symptoms of excessive anxiety in children by recognizing anxiety-specific and general problem indications. The results provided preliminary evidence that teachers’ knowledge of anxiety and anxiety disorders does not appear to be a barrier in preventing children’s referrals for mental health treatment. Implications for practice and directions for future research are discussed.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

This research makes a major contribution which enables efficient searching and indexing of large archives of spoken audio based on speaker identity. It introduces a novel technique dubbed as “speaker attribution” which is the task of automatically determining ‘who spoke when?’ in recordings and then automatically linking the unique speaker identities within each recording across multiple recordings. The outcome of the research will also have significant impact in improving the performance of automatic speech recognition systems through the extracted speaker identities.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

Context Patients with venous leg ulcers experience multiple symptoms, including pain, depression, and discomfort from lower leg inflammation and wound exudate. Some of these symptoms impair wound healing and decrease quality of life (QOL). The presence of co-occurring symptoms may have a negative effect on these outcomes. The identification of symptom clusters could potentially lead to improvements in symptom management and QOL. Objectives To identify the prevalence and severity of common symptoms and the occurrence of symptom clusters in patients with venous leg ulcers. Methods For this secondary analysis, data on sociodemographic characteristics, medical history, venous history, ulcer and lower limb clinical characteristics, symptoms, treatments, healing, and QOL were analyzed from a sample of 318 patients with venous leg ulcers who were recruited from hospital outpatient and community nursing clinics for leg ulcers. Exploratory factor analysis was used to identify symptom clusters. Results Almost two-thirds (64%) of the patients experienced four or more concurrent symptoms. The most frequent symptoms were sleep disturbance (80%), pain (74%), and lower limb swelling (67%). Sixty percent of patients reported three or more symptoms at a moderate-to-severe level of intensity (e.g., 78% reported disturbed sleep frequently or always; the mean pain severity score was 49 of 100, SD 26.5). Exploratory factor analysis identified two symptom clusters: pain, depression, sleep disturbance, and fatigue; and swelling, inflammation, exudate, and fatigue. Conclusion Two symptom clusters were identified in this sample of patients with venous leg ulcers. Further research is needed to verify these symptom clusters and to evaluate their effect on patient outcomes.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

Recent advances in the area of ‘Transformational Government’ position the citizen at the centre of focus. This paradigm shift from a department-centric to a citizen-centric focus requires governments to re-think their approach to service delivery, thereby decreasing costs and increasing citizen satisfaction. The introduction of franchises as a virtual business layer between the departments and their citizens is intended to provide a solution. Franchises are structured to address the needs of citizens independent of internal departmental structures. For delivering services online, governments pursue the development of a One-Stop Portal, which structures information and services through those franchises. Thus, each franchise can be mapped to a specific service bundle, which groups together services that are deemed to be of relevance to a specific citizen need. This study focuses on the development and evaluation of these service bundles. In particular, two research questions guide the line of investigation of this study: Research Question 1): What methods can be used by governments to identify service bundles as part of governmental One-Stop Portals? Research Question 2): How can the quality of service bundles in governmental One-Stop Portals be evaluated? The first research question asks about the identification of suitable service bundle identification methods. A literature review was conducted, to, initially, conceptualise the service bundling task, in general. As a consequence, a 4-layer model of service bundling and a morphological box were created, detailing characteristics that are of relevance when identifying service bundles. Furthermore, a literature review of Decision-Support Systems was conducted to identify approaches of relevance in different bundling scenarios. These initial findings were complemented by targeted studies of multiple leading governments in the e-government domain, as well as with a local expert in the field. Here, the aim was to identify the current status of online service delivery and service bundling in practice. These findings led to the conceptualising of two service bundle identification methods, applicable in the context of Queensland Government: On the one hand, a provider-driven approach, based on service description languages, attributes, and relationships between services was conceptualised. As well, a citizen-driven approach, based on analysing the outcomes from content identification and grouping workshops with citizens, was also conceptualised. Both methods were then applied and evaluated in practice. The conceptualisation of the provider-driven method for service bundling required the initial specification of relevant attributes that could be used to identify similarities between services called relationships; these relationships then formed the basis for the identification of service bundles. This study conceptualised and defined seven relationships, namely ‘Co-location’, ‘Resource’, ‘Co-occurrence’, ‘Event’, ‘Consumer’, ‘Provider’, and ‘Type’. The relationships, and the bundling method itself, were applied and refined as part of six Action Research cycles in collaboration with the Queensland Government. The findings show that attributes and relationships can be used effectively as a means for bundle identification, if distinct decision rules are in place to prescribe how services are to be identified. For the conceptualisation of the citizen-driven method, insights from the case studies led to the decision to involve citizens, through card sorting activities. Based on an initial list of services, relevant for a certain franchise, participating citizens grouped services according to their liking. The card sorting activity, as well as the required analysis and aggregation of the individual card sorting results, was analysed in depth as part of this study. A framework was developed that can be used as a decision-support tool to assist with the decision of what card sorting analysis method should be utilised in a given scenario. The characteristic features associated with card sorting in a government context led to the decision to utilise statistical analysis approaches, such as cluster analysis and factor analysis, to aggregate card sorting results. The second research question asks how the quality of service bundles can be assessed. An extensive literature review was conducted focussing on bundle, portal, and e-service quality. It was found that different studies use different constructs, terminology, and units of analysis, which makes comparing these models a difficult task. As a direct result, a framework was conceptualised, that can be used to position past and future studies in this research domain. Complementing the literature review, interviews conducted as part of the case studies with leaders in e-government, indicated that, typically, satisfaction is evaluated for the overall portal once the portal is online, but quality tests are not conducted during the development phase. Consequently, a research model which appropriately defines perceived service bundle quality would need to be developed from scratch. Based on existing theory, such as Theory of Reasoned Action, Expectation Confirmation Theory, and Theory of Affordances, perceived service bundle quality was defined as an inferential belief. Perceived service bundle quality was positioned within the nomological net of services. Based on the literature analysis on quality, and on the subsequent work of a focus group, the hypothesised antecedents (descriptive beliefs) of the construct and the associated question items were defined and the research model conceptualised. The model was then tested, refined, and finally validated during six Action Research cycles. Results show no significant difference in higher quality or higher satisfaction among users for either the provider-driven method or for the citizen-driven method. The decision on which method to choose, it was found, should be based on contextual factors, such as objectives, resources, and the need for visibility. The constructs of the bundle quality model were examined. While the quality of bundles identified through the citizen-centric approach could be explained through the constructs ‘Navigation’, ‘Ease of Understanding’, and ‘Organisation’, bundles identified through the provider-driven approach could be explained solely through the constructs ‘Navigation’ and ‘Ease of Understanding’. An active labelling style for bundles, as part of the provider-driven Information Architecture, had a larger impact on ‘Quality’ than the topical labelling style used in the citizen-centric Information Architecture. However, ‘Organisation’, reflecting the internal, logical structure of the Information Architecture, was a significant factor impacting on ‘Quality’ only in the citizen-driven Information Architecture. Hence, it was concluded that active labelling can compensate for a lack of logical structure. Further studies are needed to further test this conjecture. Such studies may involve building alternative models and conducting additional empirical research (e.g. use of an active labelling style for the citizen-driven Information Architecture). This thesis contributes to the body of knowledge in several ways. Firstly, it presents an empirically validated model of the factors explaining and predicting a citizen’s perception of service bundle quality. Secondly, it provides two alternative methods that can be used by governments to identify service bundles in structuring the content of a One-Stop Portal. Thirdly, this thesis provides a detailed narrative to suggest how the recent paradigm shift in the public domain, towards a citizen-centric focus, can be pursued by governments; the research methodology followed by this study can serve as an exemplar for governments seeking to achieve a citizen-centric approach to service delivery.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

Introduction: The accurate identification of tissue electron densities is of great importance for Monte Carlo (MC) dose calculations. When converting patient CT data into a voxelised format suitable for MC simulations, however, it is common to simplify the assignment of electron densities so that the complex tissues existing in the human body are categorized into a few basic types. This study examines the effects that the assignment of tissue types and the calculation of densities can have on the results of MC simulations, for the particular case of a Siemen’s Sensation 4 CT scanner located in a radiotherapy centre where QA measurements are routinely made using 11 tissue types (plus air). Methods: DOSXYZnrc phantoms are generated from CT data, using the CTCREATE user code, with the relationship between Hounsfield units (HU) and density determined via linear interpolation between a series of specified points on the ‘CT-density ramp’ (see Figure 1(a)). Tissue types are assigned according to HU ranges. Each voxel in the DOSXYZnrc phantom therefore has an electron density (electrons/cm3) defined by the product of the mass density (from the HU conversion) and the intrinsic electron density (electrons /gram) (from the material assignment), in that voxel. In this study, we consider the problems of density conversion and material identification separately: the CT-density ramp is simplified by decreasing the number of points which define it from 12 down to 8, 3 and 2; and the material-type-assignment is varied by defining the materials which comprise our test phantom (a Supertech head) as two tissues and bone, two plastics and bone, water only and (as an extreme case) lead only. The effect of these parameters on radiological thickness maps derived from simulated portal images is investigated. Results & Discussion: Increasing the degree of simplification of the CT-density ramp results in an increasing effect on the resulting radiological thickness calculated for the Supertech head phantom. For instance, defining the CT-density ramp using 8 points, instead of 12, results in a maximum radiological thickness change of 0.2 cm, whereas defining the CT-density ramp using only 2 points results in a maximum radiological thickness change of 11.2 cm. Changing the definition of the materials comprising the phantom between water and plastic and tissue results in millimetre-scale changes to the resulting radiological thickness. When the entire phantom is defined as lead, this alteration changes the calculated radiological thickness by a maximum of 9.7 cm. Evidently, the simplification of the CT-density ramp has a greater effect on the resulting radiological thickness map than does the alteration of the assignment of tissue types. Conclusions: It is possible to alter the definitions of the tissue types comprising the phantom (or patient) without substantially altering the results of simulated portal images. However, these images are very sensitive to the accurate identification of the HU-density relationship. When converting data from a patient’s CT into a MC simulation phantom, therefore, all possible care should be taken to accurately reproduce the conversion between HU and mass density, for the specific CT scanner used. Acknowledgements: This work is funded by the NHMRC, through a project grant, and supported by the Queensland University of Technology (QUT) and the Royal Brisbane and Women's Hospital (RBWH), Brisbane, Australia. The authors are grateful to the staff of the RBWH, especially Darren Cassidy, for assistance in obtaining the phantom CT data used in this study. The authors also wish to thank Cathy Hargrave, of QUT, for assistance in formatting the CT data, using the Pinnacle TPS. Computational resources and services used in this work were provided by the HPC and Research Support Group, QUT, Brisbane, Australia.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

This thesis investigated the viability of using Frequency Response Functions in combination with Artificial Neural Network technique in damage assessment of building structures. The proposed approach can help overcome some of limitations associated with previously developed vibration based methods and assist in delivering more accurate and robust damage identification results. Excellent results are obtained for damage identification of the case studies proving that the proposed approach has been developed successfully.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

Migraine is a common neurological disorder with a strong genetic basis. However, the complex nature of the disorder has meant that few genes or susceptibility loci have been identified and replicated consistently to confirm their involvement in migraine. Approaches to genetic studies of the disorder have included analysis of the rare migraine subtype, familial hemiplegic migraine with several causal genes identified for this severe subtype. However, the exact genetic contributors to the more common migraine subtypes are still to be deciphered. Genome-wide studies such as genome-wide association studies and linkage analysis as well as candidate genes studies have been employed to investigate genes involved in common migraine. Neurological, hormonal and vascular genes are all considered key factors in the pathophysiology of migraine and are a focus of many of these studies. It is clear that the influence of individual genes on the expression of this disorder will vary. Furthermore, the disorder may be dependent on gene–gene and gene–environment interactions that have not yet been considered. In addition, identifying susceptibility genes may require phenotyping methods outside of the International Classification of Headache Disorders II criteria, such as trait component analysis and latent class analysis to better define the ambit of migraine expression.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

This study investigated potential markers within chromosomal, mitochondrial DNA (mtDNA) and ribosomal RNA (rRNA) with the aim of developing a DNA based method to allow differentiation between animal species. Such discrimination tests may have important applications in the forensic science, agriculture, quarantine and customs fields. DNA samples from five different animal individuals within the same species for 10 species of animal (including human) were analysed. DNA extraction and quantitation followed by PCR amplification and GeneScan visualisation formed the basis of the experimental analysis. Five gene markers from three different types of genes were investigated. These included genomic markers for the β-actin and TP53 tumor suppressor gene. Mitochondrial DNA markers, designed by Bataille et al. [Forensic Sci. Int. 99 (1999) 165], examined the Cytochrome b gene and Hypervariable Displacement Loop (D-Loop) region. Finally, a ribosomal RNA marker for the 28S rRNA gene optimised by Naito et al. [J. Forensic Sci. 37 (1992) 396] was used as a possible marker for speciation. Results showed a difference of only several base pairs between all species for the β-actin and 28S markers, with the exception of Sus scrofa (pig) β-actin fragment length, which produced a significantly smaller fragment. Multiplexing of Cytochrome b and D-Loop markers gave limited species information, although positive discrimination of human DNA was evident. The most specific and discriminatory results were shown using the TP53 gene since this marker produced greatest fragment size differences between animal species studied. Sample differentiation for all species was possible following TP53 amplification, suggesting that this gene could be used as a potential animal species identifier.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

A high performance liquid chromatographic method for the simultaneous analysis of two flavonoids (iso-vitexin and vitexin), and three indole alkaloids (harmane, harmine, and harmol) was developed. This method was then utilised to quantitate levels of these five constituents in methanolic extracts of Australian Passiflora incarnata. HPLC analysis was performed using a Waters™ Novapak C18 (150 × 4 mm, 4 μm) column, with a gradient solvent system of methanol-water-acetic acid. Detection was achieved by PDA UV (254 nm) and fluorescence (excitation 254 nm, emission 414 nm), utilising the external standard method to obtain quantification.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

Objective To determine changes in ability to identify specific vegetables and fruits, and attitudes towards vegetables and fruit, associated with the introduction of a school-based food garden. Design A 12-month intervention trial using a historical control (control n 132, intervention n 120), class-based, self-administered questionnaires requiring one-word answers and 3-point Likert scale responses. Setting A state primary school (grades 4 to 7) in a low socio-economic area of Brisbane, Australia. Intervention The introduction of a school-based food garden, including the funding of a teacher coordinator for 11 h/week to facilitate integration of garden activities into the curriculum. Main outcome measures Ability to identify a series of vegetables and fruits, attitudes towards vegetables and fruit. Analysis Frequency distributions for each item were generated and χ2 analyses were used to determine statistical significance. Exploratory factor analysis was employed to detect major trends in data. Results The intervention led to enhanced ability to identify individual vegetables and fruits, greater attention to origins of produce (garden-grown and fresh), changes to perceived consumption of vegetables and fruits, and enhanced confidence in preparing fruit and vegetable snacks, but decreased interest in trying new fruits. Conclusions The introduction of this school-based food garden was associated with skill and attitudinal changes conducive to enhancing vegetable and fruit consumption. The ways in which such changes might impact on dietary behaviours and intake require further analysis.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

This research has successfully applied super-resolution and multiple modality fusion techniques to address the major challenges of human identification at a distance using face and iris. The outcome of the research is useful for security applications.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

Aerosol mass spectrometers (AMS) are powerful tools in the analysis of the chemical composition of airborne particles, particularly organic aerosols which are gaining increasing attention. However, the advantages of AMS in providing on-line data can be outweighed by the difficulties involved in its use in field measurements at multiple sites. In contrast to the on-line measurement by AMS, a method which involves sample collection on filters followed by subsequent analysis by AMS could significantly broaden the scope of AMS application. We report the application of such an approach to field studies at multiple sites. An AMS was deployed at 5 urban schools to determine the sources of the organic aerosols at the schools directly. PM1 aerosols were also collected on filters at these and 20 other urban schools. The filters were extracted with water and the extract run through a nebulizer to generate the aerosols, which were analysed by an AMS. The mass spectra from the samples collected on filters at the 5 schools were found to have excellent correlations with those obtained directly by AMS, with r2 ranging from 0.89 to 0.98. Filter recoveries varied between the schools from 40 -115%, possibly indicating that this method provides qualitative rather than quantitative information. The stability of the organic aerosols on Teflon filters was demonstrated by analysing samples stored for up to two years. Application of the procedure to the remaining 20 schools showed that secondary organic aerosols were the main source of aerosols at the majority of the schools. Overall, this procedure provides accurate representation of the mass spectra of ambient organic aerosols and could facilitate rapid data acquisition at multiple sites where AMS could not be deployed for logistical reasons.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

A significant amount of speech is typically required for speaker verification system development and evaluation, especially in the presence of large intersession variability. This paper introduces a source and utterance duration normalized linear discriminant analysis (SUN-LDA) approaches to compensate session variability in short-utterance i-vector speaker verification systems. Two variations of SUN-LDA are proposed where normalization techniques are used to capture source variation from both short and full-length development i-vectors, one based upon pooling (SUN-LDA-pooled) and the other on concatenation (SUN-LDA-concat) across the duration and source-dependent session variation. Both the SUN-LDA-pooled and SUN-LDA-concat techniques are shown to provide improvement over traditional LDA on NIST 08 truncated 10sec-10sec evaluation conditions, with the highest improvement obtained with the SUN-LDA-concat technique achieving a relative improvement of 8% in EER for mis-matched conditions and over 3% for matched conditions over traditional LDA approaches.

Relevância:

20.00% 20.00%

Publicador:

Resumo:

A significant amount of speech data is required to develop a robust speaker verification system, but it is difficult to find enough development speech to match all expected conditions. In this paper we introduce a new approach to Gaussian probabilistic linear discriminant analysis (GPLDA) to estimate reliable model parameters as a linearly weighted model taking more input from the large volume of available telephone data and smaller proportional input from limited microphone data. In comparison to a traditional pooled training approach, where the GPLDA model is trained over both telephone and microphone speech, this linear-weighted GPLDA approach is shown to provide better EER and DCF performance in microphone and mixed conditions in both the NIST 2008 and NIST 2010 evaluation corpora. Based upon these results, we believe that linear-weighted GPLDA will provide a better approach than pooled GPLDA, allowing for the further improvement of GPLDA speaker verification in conditions with limited development data.