Word boundary modelling and full covariance gaussians for Arabic Speech-to-Text systems

This paper describes recent improvements to the Cambridge Arabic Large Vocabulary Continuous Speech Recognition (LVCSR) Speech-to-Text (STT) system. It is shown that wordboundary context markers provide a powerful method to enhance graphemic systems by implicit phonetic information, improving the modelling capability of graphemic systems. In addition, a robust technique for full covariance Gaussian modelling in the Minimum Phone Error (MPE) training framework is introduced. This reduces the full covariance training to a diagonal covariance training problem, thereby solving related robustness problems. The full system results show that the combined use of these and other techniques within a multi-branch combination framework reduces the Word Error Rate (WER) of the complete system by up to 5.9% relative. Copyright © 2011 ISCA.

Physicians in leadership: the association between medical director involvement and staff-to-patient ratios

In a hospital environment that demands a careful balance between commercial and clinical interests, the extent to which physicians are involved in hospital leadership varies greatly. This paper assesses the influence of the extent of this involvement on staff-to-patient ratios. Using data gathered from 604 hospitals across Germany, this study evidences the positive relationship between a full-time medical director (MD) or heavily involved part-time MD and a higher staff-to-patient ratio. The data allows us to control for a range of confounding variables, such as size, rural/urban location, ownership structure, and case-mix. The results contribute to the sparse body of empirical research on the effect of clinical leadership on organizational outcomes.

Syllable language models for Mandarin speech recognition: exploiting character language models.

Mandarin Chinese is based on characters which are syllabic in nature and morphological in meaning. All spoken languages have syllabiotactic rules which govern the construction of syllables and their allowed sequences. These constraints are not as restrictive as those learned from word sequences, but they can provide additional useful linguistic information. Hence, it is possible to improve speech recognition performance by appropriately combining these two types of constraints. For the Chinese language considered in this paper, character level language models (LMs) can be used as a first level approximation to allowed syllable sequences. To test this idea, word and character level n-gram LMs were trained on 2.8 billion words (equivalent to 4.3 billion characters) of texts from a wide collection of text sources. Both hypothesis and model based combination techniques were investigated to combine word and character level LMs. Significant character error rate reductions up to 7.3% relative were obtained on a state-of-the-art Mandarin Chinese broadcast audio recognition task using an adapted history dependent multi-level LM that performs a log-linearly combination of character and word level LMs. This supports the hypothesis that character or syllable sequence models are useful for improving Mandarin speech recognition performance.

Vowel normalisation: Time-domain processing of the internal dynamics of speech

Human listeners can identify vowels regardless of speaker size, although the sound waves for an adult and a child speaking the ’same’ vowel would differ enormously. The differences are mainly due to the differences in vocal tract length (VTL) and glottal pulse rate (GPR) which are both related to body size. Automatic speech recognition machines are notoriously bad at understanding children if they have been trained on the speech of an adult. In this paper, we propose that the auditory system adapts its analysis of speech sounds, dynamically and automatically to the GPR and VTL of the speaker on a syllable-to-syllable basis. We illustrate how this rapid adaptation might be performed with the aid of a computational version of the auditory image model, and we propose that an auditory preprocessor of this form would improve the robustness of speech recognisers.

Stakeholder challenges in purchasing medical devices for patient safety.

OBJECTIVE: This study identifies the stakeholders who have a role in medical device purchasing within the wider system of health-care delivery and reports on their particular challenges to promote patient safety during purchasing decisions. METHODS: Data was collected through observational work, participatory workshops, and semi-structured qualitative interviews, which were analyzed and coded. The study takes a systems-based and engineering design approach to the study. Five hospitals took part in this study, and the participants included maintenance, training, clinical end-users, finance, and risk departments. RESULTS: The main stakeholders for purchasing were identified to be staff from clinical engineering (Maintenance), device users (Clinical), device trainers (Training), and clinical governance for analyzing incidents involving devices (Risk). These stakeholders display varied characteristics in terms of interpretation of their own roles, competencies for selecting devices, awareness and use of resources for purchasing devices, and attitudes toward the purchasing process. The role of "clinical engineering" is seen by these stakeholders to be critical in mediating between training, technical, and financial stakeholders but not always recognized in practice. CONCLUSIONS: The findings show that many device purchasing decisions are tackled in isolation, which is not optimal for decisions requiring knowledge that is currently distributed among different people within different departments. The challenges expressed relate to the wider system of care and equipment management, calling for a more systemic view of purchasing for medical devices.