Rapid yet accurate speech indexing using dynamic match lattice spotting

The support for typically out-of-vocabulary query terms such as names, acronyms, and foreign words is an important requirement of many speech indexing applications. However, to date many unrestricted vocabulary indexing systems have struggled to provide a balance between good detection rate and fast query speeds. This paper presents a fast and accurate unrestricted vocabulary speech indexing technique named Dynamic Match Lattice Spotting (DMLS). The proposed method augments the conventional lattice spotting technique with dynamic sequence matching, together with a number of other novel algorithmic enhancements, to obtain a system that is capable of searching hours of speech in seconds while maintaining excellent detection performance

Digital image databases support GIS operations

Evaluates trends in the imagery built into GIS applications to supplement existing vector data of streets, boundaries, infrastructure and utilities. These include large area digital orthophotos, Landsat and SPOT data. Future developments include 3 to 5 metre pixel resolutions from satellites, 1 to 2 metres from aircraft. GPS and improved image analysis techniques will also assist in improving resolution and accuracy.

Databases at QUT

Presentation by Dr Amadeo Pugliese, QUT Business School at Managing your research data seminar, 2012

Baidu’s perfect paradox : free speech and the right to censor

China’s biggest search engine has a constitutional right to filter its search results, a US court found last month. But that’s just the start of the story. Eight New York-based pro-democracy activists sued Baidu Inc in 2011, seeking damages because Baidu prevents their work from showing up in search results. Baidu follows Chinese law that requires it to censor politically sensitive results. But in what the plaintiffs’ lawyer has dubbed a “perfect paradox”, US District Judge Jesse Furman has dismissed the challenge, explaining that to hold Baidu liable for its decisions to censor pro-democracy content would itself infringe the right to free speech.

Using police crash databases for injury prevention research - a comparison of opt-out and opt-in approaches to study recruitment

Objective: The study aimed to examine the difference in response rates between opt-out and opt-in participant recruitment in a population-based study of heavy-vehicle drivers involved in a police-attended crash. Methods: Two approaches to subject recruitment were implemented in two different states over a 14-week period and response rates for the two approaches (opt-out versus opt-in recruitment) were compared. Results: Based on the eligible and contactable drivers, the response rates were 54% for the optout group and 16% for the opt-in group. Conclusions and Implications: The opt-in recruitment strategy (which was a consequence of one jurisdiction’s interpretation of the national Privacy Act at the time) resulted in an insufficient and potentially biased sample for the purposes of conducting research into risk factors for heavy-vehicle crashes. Australia’s national Privacy Act 1988 has had a long history of inconsistent practices by state and territory government departments and ethical review committees. These inconsistencies can have profound effects on the validity of research, as shown through the significantly different response rates we reported in this study. It is hoped that a more unified interpretation of the Privacy Act across the states and territories, as proposed under the soon-to-be released Australian Privacy Principles will reduce the recruitment challenges outlined in this study.

The effect of prior visual information on recognition of speech and sounds

To identify and categorize complex stimuli such as familiar objects or speech, the human brain integrates information that is abstracted at multiple levels from its sensory inputs. Using cross-modal priming for spoken words and sounds, this functional magnetic resonance imaging study identified 3 distinct classes of visuoauditory incongruency effects: visuoauditory incongruency effects were selective for 1) spoken words in the left superior temporal sulcus (STS), 2) environmental sounds in the left angular gyrus (AG), and 3) both words and sounds in the lateral and medial prefrontal cortices (IFS/mPFC). From a cognitive perspective, these incongruency effects suggest that prior visual information influences the neural processes underlying speech and sound recognition at multiple levels, with the STS being involved in phonological, AG in semantic, and mPFC/IFS in higher conceptual processing. In terms of neural mechanisms, effective connectivity analyses (dynamic causal modeling) suggest that these incongruency effects may emerge via greater bottom-up effects from early auditory regions to intermediate multisensory integration areas (i.e., STS and AG). This is consistent with a predictive coding perspective on hierarchical Bayesian inference in the cortex where the domain of the prediction error (phonological vs. semantic) determines its regional expression (middle temporal gyrus/STS vs. AG/intraparietal sulcus).

Split leverage : attacking the confidentiality of linked databases by partitioning

This article considers the risk of disclosure in linked databases when statistical analysis of micro-data is permitted. The risk of disclosure needs to be balanced against the utility of the linked data. The current work specifically considers the disclosure risks in permitting regression analysis to be performed on linked data. A new attack based on partitioning of the database is presented.

A Likelihood-Maximizing Framework for Enhanced In-Car Speech Recognition Based on Speech Dialog System Interaction

Speech recognition in car environments has been identified as a valuable means for reducing driver distraction when operating noncritical in-car systems. Under such conditions, however, speech recognition accuracy degrades significantly, and techniques such as speech enhancement are required to improve these accuracies. Likelihood-maximizing (LIMA) frameworks optimize speech enhancement algorithms based on recognized state sequences rather than traditional signal-level criteria such as maximizing signal-to-noise ratio. LIMA frameworks typically require calibration utterances to generate optimized enhancement parameters that are used for all subsequent utterances. Under such a scheme, suboptimal recognition performance occurs in noise conditions that are significantly different from that present during the calibration session – a serious problem in rapidly changing noise environments out on the open road. In this chapter, we propose a dialog-based design that allows regular optimization iterations in order to track the ever-changing noise conditions. Experiments using Mel-filterbank noise subtraction (MFNS) are performed to determine the optimization requirements for vehicular environments and show that minimal optimization is required to improve speech recognition, avoid over-optimization, and ultimately assist with semireal-time operation. It is also shown that the proposed design is able to provide improved recognition performance over frameworks incorporating a calibration session only.

Complete-linkage clustering for voice activity detection in audio and visual speech

We propose a novel technique for conducting robust voice activity detection (VAD) in high-noise recordings. We use Gaussian mixture modeling (GMM) to train two generic models; speech and non-speech. We then score smaller segments of a given (unseen) recording against each of these GMMs to obtain two respective likelihood scores for each segment. These scores are used to compute a dissimilarity measure between pairs of segments and to carry out complete-linkage clustering of the segments into speech and non-speech clusters. We compare the accuracy of our method against state-of-the-art and standardised VAD techniques to demonstrate an absolute improvement of 15% in half-total error rate (HTER) over the best performing baseline system and across the QUT-NOISE-TIMIT database. We then apply our approach to the Audio-Visual Database of American English (AVDBAE) to demonstrate the performance of our algorithm in using visual, audio-visual or a proposed fusion of these features.

Mind what you say-general and specific mechanisms for monitoring in speech production

For most people, speech production is relatively effortless and error-free. Yet it has long been recognized that we need some type of control over what we are currently saying and what we plan to say. Precisely how we monitor our internal and external speech has been a topic of research interest for several decades. The predominant approach in psycholinguistics has assumed monitoring of both is accomplished via systems responsible for comprehending others' speech. This special topic aimed to broaden the field, firstly by examining proposals that speech production might also engage more general systems, such as those involved in action monitoring. A second aim was to examine proposals for a production-specific, internal monitor. Both aims require that we also specify the nature of the representations subject to monitoring.

Acoustic adaptation in cross database audio visual SHMM training for phonetic spoken term detection

Visual information in the form of lip movements of the speaker has been shown to improve the performance of speech recognition and search applications. In our previous work, we proposed cross database training of synchronous hidden Markov models (SHMMs) to make use of external large and publicly available audio databases in addition to the relatively small given audio visual database. In this work, the cross database training approach is improved by performing an additional audio adaptation step, which enables audio visual SHMMs to benefit from audio observations of the external audio models before adding visual modality to them. The proposed approach outperforms the baseline cross database training approach in clean and noisy environments in terms of phone recognition accuracy as well as spoken term detection (STD) accuracy.

Cross database training of audio-visual hidden Markov models for phone recognition

Speech recognition can be improved by using visual information in the form of lip movements of the speaker in addition to audio information. To date, state-of-the-art techniques for audio-visual speech recognition continue to use audio and visual data of the same database for training their models. In this paper, we present a new approach to make use of one modality of an external dataset in addition to a given audio-visual dataset. By so doing, it is possible to create more powerful models from other extensive audio-only databases and adapt them on our comparatively smaller multi-stream databases. Results show that the presented approach outperforms the widely adopted synchronous hidden Markov models (HMM) trained jointly on audio and visual data of a given audio-visual database for phone recognition by 29% relative. It also outperforms the external audio models trained on extensive external audio datasets and also internal audio models by 5.5% and 46% relative respectively. We also show that the proposed approach is beneficial in noisy environments where the audio source is affected by the environmental noise.

Multilingualism and speech-language competence in early childhood: Impact on academic and social-emotional outcomes at school

This large-scale longitudinal population study provided a rare opportunity to consider the interface between multilingualism and speech-language competence on children’s academic and social-emotional outcomes and to determine whether differences between groups at 4 to 5 years persist, deepen, or disappear with time and schooling. Four distinct groups were identified from the Kindergarten cohort of the Longitudinal Study of Australian Children (LSAC) (1) English-only + typical speech and language (n = 2,012); (2) multilingual + typical speech and language (n = 476); (3) English-only + speech and language concern (n = 643); and (4) multilingual + speech and language concern (n = 109). Two analytic approaches were used to compare these groups. First, a matched case-control design was used to randomly match multilingual children with speech and language concern (group 4, n = 109) to children in groups 1, 2, and 3 on gender, age, and family socio-economic position in a cross-sectional comparison of vocabulary, school readiness, and behavioral adjustment. Next, analyses were applied to the whole sample to determine longitudinal effects of group membership on teachers’ ratings of literacy, numeracy, and behavioral adjustment at ages 6 to 7 and 8 to 9 years. At 4 to 5 years, multilingual children with speech and language concern did equally well or better than English-only children (with or without speech and language concern) on school readiness tests but performed more poorly on measures of English vocabulary and behavior. At ages 6 to 7 and 8 to 9, the early gap between English-only and multilingual children had closed. Multilingualism was not found to contribute to differences in literacy and numeracy outcomes at school; instead, outcomes were more related to concerns about children’s speech and language in early childhood. There were no group differences for socio-emotional outcomes. Early evidence for the combined risks of multilingualism plus speech and language concern was not upheld into the school years.