Rapid joint speaker and noise compensation for robust speech recognition

For speech recognition, mismatches between training and testing for speaker and noise are normally handled separately. The work presented in this paper aims at jointly applying speaker adaptation and model-based noise compensation by embedding speaker adaptation as part of the noise mismatch function. The proposed method gives a faster and more optimum adaptation compared to compensating for these two factors separately. It is also more consistent with respect to the basic assumptions of speaker and noise adaptation. Experimental results show significant and consistent gains from the proposed method. © 2011 IEEE.

Compression techniques applied to multiple speech recognition systems

Speech recognition systems typically contain many Gaussian distributions, and hence a large number of parameters. This makes them both slow to decode speech, and large to store. Techniques have been proposed to decrease the number of parameters. One approach is to share parameters between multiple Gaussians, thus reducing the total number of parameters and allowing for shared likelihood calculation. Gaussian tying and subspace clustering are two related techniques which take this approach to system compression. These techniques can decrease the number of parameters with no noticeable drop in performance for single systems. However, multiple acoustic models are often used in real speech recognition systems. This paper considers the application of Gaussian tying and subspace compression to multiple systems. Results show that two speech recognition systems can be modelled using the same number of Gaussians as just one system, with little effect on individual system performance. Copyright © 2009 ISCA.

Integrated online speaker clustering and adaptation

For many applications, it is necessary to produce speech transcriptions in a causal fashion. To produce high quality transcripts, speaker adaptation is often used. This requires online speaker clustering and incremental adaptation techniques to be developed. This paper presents an integrated approach to online speaker clustering and adaptation which allows efficient clustering of speakers using the same accumulated statistics that are normally used for adaptation. Using a consistent criterion for both clustering and adaptation should yield gains for both stages. The proposed approach is evaluated on a meetings transcription task using audio from multiple distant microphones. Consistent gains over standard clustering and adaptation were obtained. Copyright © 2011 ISCA.

Continuous asr for flexible incremental dialogue

Spoken dialogue systems provide a convenient way for users to interact with a machine using only speech. However, they often rely on a rigid turn taking regime in which a voice activity detection (VAD) module is used to determine when the user is speaking and decide when is an appropriate time for the system to respond. This paper investigates replacing the VAD and discrete utterance recogniser of a conventional turn-taking system with a continuously operating recogniser that is always listening, and using the recogniser 1-best path to guide turn taking. In this way, a flexible framework for incremental dialogue management is possible. Experimental results show that it is possible to remove the VAD component and successfully use the recogniser best path to identify user speech, with more robustness to noise, potentially smaller latency times, and a reduction in overall recognition error rate compared to using the conventional approach. © 2013 IEEE.

On-line policy optimisation of Bayesian spoken dialogue systems via human interaction

A partially observable Markov decision process has been proposed as a dialogue model that enables robustness to speech recognition errors and automatic policy optimisation using reinforcement learning (RL). However, conventional RL algorithms require a very large number of dialogues, necessitating a user simulator. Recently, Gaussian processes have been shown to substantially speed up the optimisation, making it possible to learn directly from interaction with human users. However, early studies have been limited to very low dimensional spaces and the learning has exhibited convergence problems. Here we investigate learning from human interaction using the Bayesian Update of Dialogue State system. This dynamic Bayesian network based system has an optimisation space covering more than one hundred features, allowing a wide range of behaviours to be learned. Using an improved policy model and a more robust reward function, we show that stable learning can be achieved that significantly outperforms a simulator trained policy. © 2013 IEEE.

Modelling relative motion to facilitate intra-limb coordination

The importance of relative motion information when modelling a novel motor skill was examined. Participants were assigned to one of four groups. Groups 1 and 2 viewed demonstrations of a skilled cricket bowler presented in either video or point light format. Group 3 observed a single point of light pertaining to the wrist of the skilled bowler only. Participants in Group 4 did not receive a demonstration and acted as controls. During 60 acquisition trials, participants in the demonstration groups viewed a model five times before each 10-trial block. Retention was examined the following day. Intra-limb coordination was assessed for the right elbow relative to the wrist in comparison to the model. The demonstration groups showed greater concordance with the model than the control group. However, the wrist group performed less like the model than the point light and video groups, who did not differ from each other. These effects were maintained in retention. Relative motion information aided the acquisition of intra-limb coordination, while making this information more salient (through point lights) provided no additional benefit. The motion of the models bowling arm was replicated more closely than the non-bowling arm, suggesting that information from the end-effector is prioritized during observation for later reproduction.

Postoperative residuial block after intermediate-acting neuromuscular blocking drugs

Summary The frequency and duration of postoperative residual neuromuscular block on arrival of 150 patients in the recovery ward following the use of vecuronium (n = 50), atracurium (n = 50) and rocuronium (n = 50) were recorded. Residual block was defined as a train-of-four ratio of 0.8 after arrival in the recovery ward were 9.2 [1-61], 6.9 [1-24] and 14.7 [1.5-83] min for the vecuronium, atracurium and rocuronium, respectively. None of the 10 patients who did not receive neuromuscular blocking drugs had train-of-four ratios

Manual versus target-controlled infusions of propofol

Summary Target-controlled infusion systems have been shown to result in the administration of larger doses of propofol, which may result in delayed emergence and recovery from anaesthesia. The aim of this study was to investigate if this was due to a difference in the depth of hypnosis (using the bispectral index monitoring) between the manual and target controlled systems of administration. Fifty unpremedicated patients undergoing elective surgery were randomly allocated to have their anaesthesia maintained with manual or target-controlled propofol infusion schemes. In both groups, the rate of propofol administration was adjusted according to the standard clinical criteria while bispectral index scores were recorded by an observer not involved in the delivery of anaesthesia. The total dose of propofol used was higher in the target controlled group (mean 9.9 [standard deviation 1.6] compared with 8.1 [1.0] mg.kg.h in the manual group [p

Neostigmine antagonism of rocuronium block during anesthesia with sevoflurane, isoflurane or propofol

Purpose: To examine the influence of continuing administration of sevoflurane or isoflurane during reversal of rocuronium induced neuromuscular block with neostigmine. Methods: One hundred and twenty patients, divided into three equal groups, were randomly allocated to maintenance of anesthesia with sevoflurane, isoflurane or propofol. Neuromuscular block was induced with rocuronium and monitored using train-of-four (TOF) stimulation of the ulnar nerve and recording the force of contraction of the adductor pollicis muscle. Neostigmine was administered when the first response in TOF had recovered to 25%. At this time the volatile agent administration was stopped or propofol dosage reduced in half the patients in each group (n = 20 in each group). The times to attain TOF ratio of 0.8, and the number of patients attaining this end point within 15 min were recorded. Results: The times (mean ± SD) to recovery of the TOF ratio to 0.8 were 12.0 ± 5.5 and 6.8 ± 2.3 min in the sevoflurane continued and sevoflurane stopped groups, 9.0 ± 8.3 and 5.5 ± 3.0 min in the isoflurane continued and isoflurane stopped groups, and 5.2 ± 2.8 and 4.7 ±1.5 min in the propofol continued and propofol stopped groups (P <0.5- 01). Only 9 and 15 patients in the sevoflurane and isoflurane continued groups respectively had attained a TOF ratio of 0.8 within 15 min (P <0.001 for sevoflurane). Conclusions: The continued administration of sevoflurane, and to a smaller extent isoflurane, results in delay in attaining adequate antagonism of rocuronium induced neuromuscular block.