Minimising speaker verification utterance length through confidence based early verification decisions

This paper presents a novel approach of estimating the confidence interval of speaker verification scores. This approach is utilised to minimise the utterance lengths required in order to produce a confident verification decision. The confidence estimation method is also extended to address both the problem of high correlation in consecutive frame scores, and robustness with very limited training samples. The proposed technique achieves a drastic reduction in the typical data requirements for producing confident decisions in an automatic speaker verification system. When evaluated on the NIST 2005 SRE, the early verification decision method demonstrates that an average of 5–10 seconds of speech is sufficient to produce verification rates approaching those achieved previously using an average in excess of 100 seconds of speech.

Improving short utterance based I-vector speaker recognition using source and utterance-duration normalization techniques

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.

Improving short utterance i-vector speaker verification using utterance variance modelling and compensation techniques

This paper proposes techniques to improve the performance of i-vector based speaker verification systems when only short utterances are available. Short-length utterance i-vectors vary with speaker, session variations, and the phonetic content of the utterance. Well established methods such as linear discriminant analysis (LDA), source-normalized LDA (SN-LDA) and within-class covariance normalisation (WCCN) exist for compensating the session variation but we have identified the variability introduced by phonetic content due to utterance variation as an additional source of degradation when short-duration utterances are used. To compensate for utterance variations in short i-vector speaker verification systems using cosine similarity scoring (CSS), we have introduced a short utterance variance normalization (SUVN) technique and a short utterance variance (SUV) modelling approach at the i-vector feature level. A combination of SUVN with LDA and SN-LDA is proposed to compensate the session and utterance variations and is shown to provide improvement in performance over the traditional approach of using LDA and/or SN-LDA followed by WCCN. An alternative approach is also introduced using probabilistic linear discriminant analysis (PLDA) approach to directly model the SUV. The combination of SUVN, LDA and SN-LDA followed by SUV PLDA modelling provides an improvement over the baseline PLDA approach. We also show that for this combination of techniques, the utterance variation information needs to be artificially added to full-length i-vectors for PLDA modelling.

Short utterance PLDA speaker verification using SN-WLDA and variance modelling techniques

This paper proposes a combination of source-normalized weighted linear discriminant analysis (SN-WLDA) and short utterance variance (SUV) PLDA modelling to improve the short utterance PLDA speaker verification. As short-length utterance i-vectors vary with the speaker, session variations and phonetic content of the utterance (utterance variation), a combined approach of SN-WLDA projection and SUV PLDA modelling is used to compensate the session and utterance variations. Experimental studies have found that a combination of SN-WLDA and SUV PLDA modelling approach shows an improvement over baseline system (WCCN[LDA]-projected Gaussian PLDA (GPLDA)) as this approach effectively compensates the session and utterance variations.

'We just make the pictures...?' How work is portrayed in children's feature length films

This article explores how adult paid work is portrayed in 'family' feature length films. The study extends previous critical media literature which has overwhelmingly focused on depictions of gender and violence, exploring the visual content of films that is relevant to adult employment. Forty-two G/PG films were analyzed for relevant themes. Consistent with the exploratory nature of the research, themes emerged inductively from the films' content. Results reveal six major themes: males are more visible in adult work roles than women; the division of labour remains gendered; work and home are not mutually exclusive domains; organizational authority and power is wielded in punitive ways; there are avenues to better employment prospects; and status/money is paramount. The findings of the study reflect a range of subject matters related to occupational characteristics and work-related communication and interactions which are typically viewed by children in contemporary society.

Using Trauma Injury Severity Score (TRISS) variables to predict length of hospital stay following trauma in New Zealand

Aim – To develop and assess the predictive capabilities of a statistical model that relates routinely collected Trauma Injury Severity Score (TRISS) variables to length of hospital stay (LOS) in survivors of traumatic injury. Method – Retrospective cohort study of adults who sustained a serious traumatic injury, and who survived until discharge from Auckland City, Middlemore, Waikato, or North Shore Hospitals between 2002 and 2006. Cubic-root transformed LOS was analysed using two-level mixed-effects regression models. Results – 1498 eligible patients were identified, 1446 (97%) injured from a blunt mechanism and 52 (3%) from a penetrating mechanism. For blunt mechanism trauma, 1096 (76%) were male, average age was 37 years (range: 15-94 years), and LOS and TRISS score information was available for 1362 patients. Spearman’s correlation and the median absolute prediction error between LOS and the original TRISS model was ρ=0.31 and 10.8 days, respectively, and between LOS and the final multivariable two-level mixed-effects regression model was ρ=0.38 and 6.0 days, respectively. Insufficient data were available for the analysis of penetrating mechanism models. Conclusions – Neither the original TRISS model nor the refined model has sufficient ability to accurately or reliably predict LOS. Additional predictor variables for LOS and other indicators for morbidity need to be considered.

Using a longitudinal model to estimate the effect of methicillin-resistant staphylococcus aureus infection on length of stay in an intensive care unit

Healthcare-associated methicillin-resistant Staphylococcus aureus(MRSA) infection may cause increased hospital stay or, sometimes, death. Quantifying this effect is complicated because it is a time-dependent exposure: infection may prolong hospital stay, while longer stays increase the risk of infection. We overcome these problems by using a multinomial longitudinal model for estimating the daily probability of death and discharge. We then extend the basic model to estimate how the effect of MRSA infection varies over time, and to quantify the number of excess ICU days due to infection. We find that infection decreases the relative risk of discharge (relative risk ratio = 0.68, 95% credible interval: 0.54, 0.82), but is only indirectly associated with increased mortality. An infection on the first day of admission resulted in a mean extra stay of 0.3 days (95% CI: 0.1, 0.5) for a patient with an APACHE II score of 10, and 1.2 days (95% CI: 0.5, 2.0) for a patient with an APACHE II score of 30. The decrease in the relative risk of discharge remained fairly constant with day of MRSA infection, but was slightly stronger closer to the start of infection. These results confirm the importance of MRSA infection in increasing ICU stay, but suggest that previous work may have systematically overestimated the effect size.

A comparison of different methods to determine diffusion length in dye-sensitized solar cells

A new steady state method for determination of the electron diffusion length in dye-sensitized solar cells (DSCs) is described and illustrated with data obtained using cells containing three different types of electrolyte. The method is based on using near-IR absorbance methods to establish pairs of illumination intensity for which the total number of trapped electrons is the same at open circuit (where all electrons are lost by interfacial electron transfer) as at short circuit (where the majority of electrons are collected at the contact). Electron diffusion length values obtained by this method are compared with values derived by intensity modulated methods and by impedance measurements under illumination. The results indicate that the values of electron diffusion length derived from the steady state measurements are consistently lower than the values obtained by the non steady-state methods. For all three electrolytes used in the study, the electron diffusion length was sufficiently high to guarantee electron collection efficiencies greater than 90%. Measurement of the trap distributions by near-IR absorption confirmed earlier observations of much higher electron trap densities for electrolytes containing Li+ ions. It is suggested that the electron trap distributions may not be intrinsic properties of the TiO2 nanoparticles, but may be associated with electron-ion interactions.

Is length of experience an appropriate criterion to identify level of expertise?

Clinical experience plays an important role in the development of expertise, particularly when coupled with reflection on practice. There is debate, however, regarding the amount of clinical experience that is required to become an expert. Various lengths of practice have been suggested as suitable for determining expertise, ranging from five years to 15 years. This study aimed to investigate the association between length of experience and therapists’ level of expertise in the field of cerebral palsy with upper limb hypertonicity using an empirical procedure named Cochrane–Weiss–Shanteau (CWS). The methodology involved re-analysis of quantitative data collected in two previous studies. In Study 1, 18 experienced occupational therapists made hypothetical clinical decisions related to 110 case vignettes, while in Study 2, 29 therapists considered 60 case vignettes drawn randomly from those used in Study 1. A CWS index was calculated for each participant's case decisions. Then, in each study, Spearman's rho was calculated to identify the correlations between the duration of experience and level of expertise. There was no significant association between these two variables in both studies. These analyses corroborated previous findings of no association between length of experience and judgemental performance. Therefore, length of experience may not be an appropriate criterion for determining level of expertise in relation to cerebral palsy practice.

Automatic spoken language identification utilizing acoustic and phonetic speech information

Automatic spoken Language Identi¯cation (LID) is the process of identifying the language spoken within an utterance. The challenge that this task presents is that no prior information is available indicating the content of the utterance or the identity of the speaker. The trend of globalization and the pervasive popularity of the Internet will amplify the need for the capabilities spoken language identi¯ca- tion systems provide. A prominent application arises in call centers dealing with speakers speaking di®erent languages. Another important application is to index or search huge speech data archives and corpora that contain multiple languages. The aim of this research is to develop techniques targeted at producing a fast and more accurate automatic spoken LID system compared to the previous National Institute of Standards and Technology (NIST) Language Recognition Evaluation. Acoustic and phonetic speech information are targeted as the most suitable fea- tures for representing the characteristics of a language. To model the acoustic speech features a Gaussian Mixture Model based approach is employed. Pho- netic speech information is extracted using existing speech recognition technol- ogy. Various techniques to improve LID accuracy are also studied. One approach examined is the employment of Vocal Tract Length Normalization to reduce the speech variation caused by di®erent speakers. A linear data fusion technique is adopted to combine the various aspects of information extracted from speech. As a result of this research, a LID system was implemented and presented for evaluation in the 2003 Language Recognition Evaluation conducted by the NIST.