The Experts below are selected from a list of 429237 Experts worldwide ranked by ideXlab platform
Steve J. Young - One of the best experts on this subject based on the ideXlab platform.
-
Robust continuous speech recognition using Parallel Model combination
IEEE Transactions on Speech and Audio Processing, 1996Co-Authors: Mark John Francis Gales, Steve J. YoungAbstract:This paper addresses the problem of automatic speech recognition in the presence of interfering noise. It focuses on the Parallel Model combination (PMC) scheme, which has been shown to be a powerful technique for achieving noise robustness. Most experiments reported on PMC to date have been on small, 10-50 word vocabulary systems. Experiments on the Resource Management (RM) database, a 1000 word continuous speech recognition task, reveal compensation requirements not highlighted by the smaller vocabulary tasks. In particular, that it is necessary to compensate the dynamic parameters as well as the static parameters to achieve good recognition performance. The database used for these experiments was the RM speaker independent task with either Lynx Helicopter noise or Operation Room noise from the NOISEX-92 database added. The experiments reported here used the HTK RM recognizer developed at CUED modified to include PMC based compensation for the static, delta and delta-delta parameters. After training on clean speech data, the performance of the recognizer was found to be severely degraded when noise was added to the speech signal at between 10 and 18 dB. However, using PMC the performance was restored to a level comparable with that obtained when training directly in the noise corrupted environment
Steve Young - One of the best experts on this subject based on the ideXlab platform.
-
robust speech recognition in additive and convolutional noise using Parallel Model combination
Computer Speech & Language, 1995Co-Authors: Mark John Francis Gales, Steve YoungAbstract:Abstract The method of Parallel Model Combination (PMC) has been shown to be a powerful technique for compensating a speech recognizer for the effects of additive noise. In this paper, the PMC scheme is extended to include the effects of convolutional noise. This is done by introducing a modified “mismatch” function which allows an estimate to be made of the difference in channel conditions or tilt between training and test environments. Having estimated this tilt, Maximum Likelihood (ML) estimates of the corrupted speech Model may then be obtained in the usual way. The scheme is evaluated using the NOISEX-92 database where the performance in the presence of both interfering additive noise and convolutional noise shows only slight degradation compared with that obtained when no convolutional noise is present.
-
robust speech recognition in additive and convolutional noise using Parallel Model combination
Computer Speech & Language, 1995Co-Authors: Mark John Francis Gales, Steve YoungAbstract:Abstract The method of Parallel Model Combination (PMC) has been shown to be a powerful technique for compensating a speech recognizer for the effects of additive noise. In this paper, the PMC scheme is extended to include the effects of convolutional noise. This is done by introducing a modified “mismatch” function which allows an estimate to be made of the difference in channel conditions or tilt between training and test environments. Having estimated this tilt, Maximum Likelihood (ML) estimates of the corrupted speech Model may then be obtained in the usual way. The scheme is evaluated using the NOISEX-92 database where the performance in the presence of both interfering additive noise and convolutional noise shows only slight degradation compared with that obtained when no convolutional noise is present.
-
ICASSP - A fast and flexible implementation of Parallel Model combination
1995 International Conference on Acoustics Speech and Signal Processing, 1Co-Authors: Mark J. F. Gales, Steve YoungAbstract:In previous papers the use of Parallel Model combination (PMC) for noise robustness has been described. Various fast implementations have been proposed, though to date in order to compensate all the parameters of a system it has been necessary to perform Gaussian integration. This paper introduces an alternative method that can compensate all the parameters of the recognition system, whilst reducing the computational load of this task. Furthermore, the technique offers an additional degree of flexibility, as it allows the number of components to be chosen and optimised using standard iterative techniques. The new technique is referred to as data-driven PMC (DPMC). It is evaluated on the Resource Management database, with noise artificially added from the NOISEX-92 database. The performance of DPMC is found to be comparable to PMC, at a far lower computational cost. In complex noise environments, by more accurately Modelling the noise source, using multiple components, and then reducing the number of components to the original number a slight improvement in performance is obtained.
Mark John Francis Gales - One of the best experts on this subject based on the ideXlab platform.
-
Robust continuous speech recognition using Parallel Model combination
IEEE Transactions on Speech and Audio Processing, 1996Co-Authors: Mark John Francis Gales, Steve J. YoungAbstract:This paper addresses the problem of automatic speech recognition in the presence of interfering noise. It focuses on the Parallel Model combination (PMC) scheme, which has been shown to be a powerful technique for achieving noise robustness. Most experiments reported on PMC to date have been on small, 10-50 word vocabulary systems. Experiments on the Resource Management (RM) database, a 1000 word continuous speech recognition task, reveal compensation requirements not highlighted by the smaller vocabulary tasks. In particular, that it is necessary to compensate the dynamic parameters as well as the static parameters to achieve good recognition performance. The database used for these experiments was the RM speaker independent task with either Lynx Helicopter noise or Operation Room noise from the NOISEX-92 database added. The experiments reported here used the HTK RM recognizer developed at CUED modified to include PMC based compensation for the static, delta and delta-delta parameters. After training on clean speech data, the performance of the recognizer was found to be severely degraded when noise was added to the speech signal at between 10 and 18 dB. However, using PMC the performance was restored to a level comparable with that obtained when training directly in the noise corrupted environment
-
robust speech recognition in additive and convolutional noise using Parallel Model combination
Computer Speech & Language, 1995Co-Authors: Mark John Francis Gales, Steve YoungAbstract:Abstract The method of Parallel Model Combination (PMC) has been shown to be a powerful technique for compensating a speech recognizer for the effects of additive noise. In this paper, the PMC scheme is extended to include the effects of convolutional noise. This is done by introducing a modified “mismatch” function which allows an estimate to be made of the difference in channel conditions or tilt between training and test environments. Having estimated this tilt, Maximum Likelihood (ML) estimates of the corrupted speech Model may then be obtained in the usual way. The scheme is evaluated using the NOISEX-92 database where the performance in the presence of both interfering additive noise and convolutional noise shows only slight degradation compared with that obtained when no convolutional noise is present.
-
robust speech recognition in additive and convolutional noise using Parallel Model combination
Computer Speech & Language, 1995Co-Authors: Mark John Francis Gales, Steve YoungAbstract:Abstract The method of Parallel Model Combination (PMC) has been shown to be a powerful technique for compensating a speech recognizer for the effects of additive noise. In this paper, the PMC scheme is extended to include the effects of convolutional noise. This is done by introducing a modified “mismatch” function which allows an estimate to be made of the difference in channel conditions or tilt between training and test environments. Having estimated this tilt, Maximum Likelihood (ML) estimates of the corrupted speech Model may then be obtained in the usual way. The scheme is evaluated using the NOISEX-92 database where the performance in the presence of both interfering additive noise and convolutional noise shows only slight degradation compared with that obtained when no convolutional noise is present.
Wei Chen - One of the best experts on this subject based on the ideXlab platform.
-
using heuristic value prediction and dynamic task granularity resizing to improve software speculation
The Scientific World Journal, 2014Co-Authors: Fan Xu, Bo Su, Zhiying Wang, Li Shen, Wei ChenAbstract:Exploiting potential thread-level Parallelism (TLP) is becoming the key factor to improving performance of programs on multicore or many-core systems. Among various kinds of Parallel execution Models, the software-based speculative Parallel Model has become a research focus due to its low cost, high efficiency, flexibility, and scalability. The performance of the guest program under the software-based speculative Parallel execution Model is closely related to the speculation accuracy, the control overhead, and the rollback overhead of the Model. In this paper, we first analyzed the conventional speculative Parallel Model and presented an analytic Model of its expectation of the overall overhead, then optimized the conventional Model based on the analytic Model, and finally proposed a novel speculative Parallel Model named HEUSPEC. The HEUSPEC Model includes three key techniques, namely, the heuristic value prediction, the value based correctness checking, and the dynamic task granularity resizing. We have implemented the runtime system of the Model in ANSI C language. The experiment results show that when the speedup of the HEUSPEC Model can reach 2.20 on the average (15% higher than conventional Model) when depth is equal to 3 and 4.51 on the average (12% higher than conventional Model) when speculative depth is equal to 7. Besides, it shows good scalability and lower memory cost.
-
ICPP - HEUSPEC: A Software Speculation Parallel Model
2013 42nd International Conference on Parallel Processing, 2013Co-Authors: Li Shen, Zhiying Wang, Hui Guo, Wei ChenAbstract:Conventional software speculative Parallel Models are facing challenge due to the increasing number of the processor core and the diversification of the application. The performance of the guest program under the software speculative Parallel execution Model is closely related to the speculation accuracy, the control overhead and the rollback overhead of the Model. In order to improve the speculative accuracy and the load balance, as well as improve the overhead of the conventional Model, in this paper, we proposed a novel speculative Parallel Model named HEUSPEC. The HEUSPEC includes 2 key techniques, the heuristic value prediction(HVP) and the dynamic task granularity resizing(DTGR). We have implemented the runtime system of the Model in ANSI C language. The experiment results show that when the speedup of the HEUSPEC Model can reach 4.51 on the average (12% higher than conventional Model) when speculative depth equals to 7. Besides, it shows good scalability and lower memory cost.
Kenneth B Sloan - One of the best experts on this subject based on the ideXlab platform.
-
flux through silicone and human skin fitted to a series Parallel Model
Therapeutic Delivery, 2014Co-Authors: John Prybylski, Kenneth B SloanAbstract:Background: Recent reports of the good correlation between maximum flux through human skin in vitro from water, JMHAQ, and maximum flux through silicone from water, JMPAQ, demand that the mechanism of maximum flux across these two apparently quite different membranes be compared to understand the bases of the correlation. Results/discussion: A n = 70 log JMPAQ database and a matched n = 55 log JMHAQ database of molecules were found to fit well to a series/Parallel Model where three Parallel solubility dependent pathways existed: a lipid pathway, an aqueous pathway, and a series pathway of alternating lipid and aqueous phases. Conclusion: The results of this analysis surprisingly suggest that the architecture of the two membranes present similar solubility based pathways through which drugs diffuse.
-
Prediction of transdermal flux of prodrugs of 5-fluorouracil, theophylline, and 6-mercaptopurine with a series/Parallel Model.
Journal of pharmaceutical sciences, 2000Co-Authors: William J. Roberts, Kenneth B SloanAbstract:Abstract Multiple regression analysis of fluxes from suspensions in isopropyl myristate ( J M ) as a function of molecular weights (MW) and solubilities in isopropyl myristate ( S IPM ) and water ( S AQ ) were performed on a data set of 41 compounds ( n = 41) comprising 39 prodrugs of 5‐fluorouracil (5‐FU), theophylline (Th), and 6‐mercaptopurine (6‐MP), in addition to 5‐FU and Th, using four Models. Two series/Parallel Models have been developed that allow an aqueous‐only path in Parallel with a lipid‐only path and with a lipid–aqueous series path for the permeation of solutes through skin: log J M = log {1/[1/( aS LIPID 10 ΦMW ) + 1/( bS AQ /MW 1/2 )] + cS LIPID 10 ΦMW + dS AQ /MW 1/2 } where a, b, c, and d are coefficients for flux through the lipid and aqueous portions of the series path, the lipid‐only path, and the aqueous‐only path, respectively, and Φ is the dependence of diffusivity in lipid on MW. In the first series/Parallel Model, S LIPID was predicted by S IPM , and in the second, solvatochromic series/Parallel Model, S LIPID was predicted by S IPM k MW + Ω i where Ω i is the sum of the solvatochromic terms α, β, π, and R 2 , and k is the coefficient for the dependence of partitioning on MW. Using the n = 41 solutions, the coefficients for the aqueous‐only path were very small or not different from zero in the two series/Parallel Models, so only two‐path series/Parallel Models were compared with the solvatochromic and transformed Potts–Guy Models where a homogeneous barrier to permeation was assumed. For each Model, one compound at a time was omitted from the data set and new parameter estimates were obtained for these 41−1 solutions and used to predict log J M for the omitted compound. The average errors of prediction of log J M (experimental log J M − predicted log J M ) for the four Models were 0.134 for the series/Parallel ( r 2 = 0.937), 0.127 for the solvatochromic series/Parallel ( r 2 = 0.967), 0.150 for the solvatochromic ( r 2 = 0.950), and 0.134 log units for the transformed Potts–Guy Model ( r 2 = 0.944). Thus, the solvatochromic series/Parallel Model provides fit and predictive ability comparable to or slightly superior to previous Models that assumed homogeneity of the diffusional barrier to flux in the rate‐determining step, provides further theoretical support against the existence of a high capacity aqueous‐only path, and provides further insight into the physicochemical properties that should be incorporated into solutes to optimize their flux. Using the solvatochromic series/Parallel Model, the parameter estimates for the n = 41 solution were used to calculate the flux of each compound through the two paths. For compounds with log partition coefficients ( K IPM:AQ ) of K IPM:AQ >1.0, permeation was mostly by the lipid‐only path; the lipid‐aqueous series path exhibited the higher carrying capacity. © 2000 Wiley‐Liss, Inc. and the American Pharmaceutical Association J Pharm Sci 89: 1415–1431, 2000
