THE CU-HTK MARCH 2000 HUB5E TRANSCRIPTION SYSTEM
Published 19 May 2000
Thomas Hain, Philip C. Woodland, G. Evermann, Daniel Povey
Citations38
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
A range of new features have been added to the HTK system used in the 1998 Hub5 evaluation, and the changes taken together have resulted in an 11% relative decrease in word error rate on the 1998 evaluation test set.
Abstract
This paper describes the Cambridge University HTK (CU-HTK) system developed for the NIST March 2000 evaluation of English conversational telephone speech transcription (Hub5E). A range of new features have been added to the HTK system used in the 1998 Hub5 evaluation, and the changes taken together have resulted in an 11% relative decrease in word error rate on the 1998 evaluation test set. Major changes include the use of maximum mutual information estimation in training as well as conventional maximum likelihood estimation; the use of a full variance transform for adaptation; the inclusion of unigram pronunciation probabilities; and word-level posterior probability estimation using confusion networks for use in minimum word error rate decoding, confidence score estimation and system combination. On the March 2000 Hub5 evaluation set the CU-HTK system gave an overall word error rate of 25.4%, which was the best performance by a statistically significant margin. This paper describes th...
Keywords
Computer Science
Computer Speech & LanguageMaximum likelihood linear regression for speaker adaptation of continuous density hidden Markov models
2,210 Citations1995C.J. Leggetter, Philip C. Woodland
An important feature of the method is that arbitrary adaptation data can be used—no special enrolment sentences are needed and that as more data is used the adaptation performance improves.
Computer Speech & LanguageMaximum likelihood linear transformations for HMM-based speech recognition
1,490 Citations1998Mark Gales
The paper compares the two possible forms of model-based transforms: unconstrained, where any combination of mean and variance transform may be used, and constrained, which requires the variance transform to have the same form as the mean transform.
A post-processing system to yield reduced word error rates: Recognizer Output Voting Error Reduction (ROVER)
1,093 Citations2002J. Fiscus
A post-recognition process which models the output generated by multiple ASR systems as independent knowledge sources that can be combined and used to generate an output with reduced error rate.
Tree-based state tying for high accuracy acoustic modelling
668 Citations1994S.J. Young, J. J. Odell +1 more
This paper describes a method of creating a tied-state continuous speech recognition system using a phonetic decision tree, which is shown to lead to similar recognition performance to that obtained using an earlier data-driven approach but to have the additional advantage of providing a mapping for unseen triphones.
Computer Speech & LanguageMean and variance adaptation within the MLLR framework
396 Citations1996Mark Gales, Philip C. Woodland
This paper examines the maximum likelihood linear regression (MLLR) adaptation technique, which has been applied to the mean parameters in mixture-Gaussian HMM systems and is extended to also update the Gaussian variances and re-estimation formulae are derived for these variance transforms.
Posterior probability decoding, confidence estimation and system combination
234 Citations2000G. Evermann, PC Woodland
The word lattices produced by the Viterbi decoder were used to generate confusion networks, which provide a compact representation of the most likely word hypotheses and their associated word posterior probabilities.
Finding consensus among words: lattice-based word error minimization
217 Citations1999Lidia Mangu, Eric Brill +1 more
A new algorithm for finding the hypothesis in a recognition lattice that is expected to minimize the word error rate (WER) is described, which overcomes the mismatch between the word-based performance metric and the standard MAP scoring paradigm that is sentence-based.
IEEE Transactions on Information TheoryAn inequality for rational functions with applications to some statistical estimation problems
198 Citations1991P.S. Gopalakrishnan, Dimitri Kanevsky +2 more
The well-known Baum-Eagon inequality provides an effective iterative scheme for finding a local maximum for homogeneous polynomials with positive coefficients over a domain of probability values.
Speech CommunicationMMIE training of large vocabulary recognition systems
177 Citations1997V. Valtchev, J. J. Odell +2 more
Experimental results show that MMIE optimisation of system structure and parameters can yield useful increases in recognition accuracy and the use of lattices makes MMIe training practicable for very complex recognition systems and large training sets.
The 1998 HTK system for transcription of conversational telephone speech
78 Citations1999Thomas Hain, Philip C. Woodland +2 more
This paper describes the 1998 HTK large vocabulary speech recognition system for conversational telephone speech as used in the NIST 1998 Hub5E evaluation and experimental results for each stage of the multi-pass decoding scheme are presented.
An improved MMIE training algorithm for speaker-independent, small vocabulary, continuous speech recognition
73 Citations1991Yves Normandin, S.D. Morgera
A corrective MMIE training algorithm is introduced, which, when applied to the TI/NIST connected digit database, has made it possible to reduce the string error rate by close to 50%.
Comparison of part-of-speech and automatically derived category-based language models for speech recognition
67 Citations2002Thomas Niesler, E. W. D. Whittaker +1 more
This paper compares various category-based language models when used in conjunction with a word-based trigram by means of linear interpolation to find the largest improvement with a model using automatically determined categories.
Broadcast news transcription using HTK
61 Citations2002Philip C. Woodland, Mark Gales +2 more
The issues in extending a large vocabulary speech recognition system designed for clean and noisy read speech tasks to handle broadcast news transcription are examined and a number of new features over previous HTK large vocabulary systems are described.
Iterative unsupervised adaptation using maximum likelihood linear regression
50 Citations2002Philip C. Woodland, David Pye +1 more
This paper investigates the iterative use ofMaximum likelihood linear regression in the context of large-vocabulary speaker-independent transcription of both noise-free and noisy data and shows that iterative application of MLLR can be beneficial especially in situations of severe mismatch.
Frame discrimination training for HMMs for large vocabulary speech recognition
36 Citations1999Daniel Povey, Philip C. Woodland
Experiments on the resource management and North American business tasks show that FD training can give comparable improvements to MMI, but is less computationally intensive.
Probabilistic classification of HMM states for large vocabulary continuous speech recognition
31 Citations1999Xiaoqiang Luo, F. Jelinek
A probabilistic classification of HMM states (PCHMM) is proposed and a distribution from a HMM state to classes is introduced, which makes the acoustic model more robust against the possible mismatch or variation between training and test data.
Large scale MMIE training for conversational telephone speech recognition
31 Citations2000PC Woodland, Daniel Povey
A lattice-based framework for maximum mutual information estimation (MMIE) of HMM parameters which has been used to train HMM systems for conversational telephone speech transcription using up to 265 hours of training data is described.