A comparison of several approximate algorithms for finding multiple (N-best) sentence hypotheses

Schwartz, Richard; Austin, S.

doi:10.1109/icassp.1991.150436

Cited by 110 publications

(50 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Even if the spoken word sequence is pruned, its time alignment will be included, before FB-wordprobabilities are calculated. Both in small and large vocabulary applications, word graphs were produced using the word pair approximation (Schwartz and Austin, 1991;Ortmanns et al, 1997). Using word probabilities and Viterbi alignment, the generalized FB probability simpli®es to the following expression:…”

Section: Discriminative Training Using Word Graphsmentioning

confidence: 99%

Comparison of discriminative training criteria

Schlüter

Macherey²

Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '98 (Cat. No.98CH36181

View full text Add to dashboard Cite

The aim of this work is to build up a common framework for a class of discriminative training criteria and optimization methods for continuous speech recognition. A uni®ed discriminative criterion based on likelihood ratios of correct and competing models with optional smoothing is presented. The uni®ed criterion leads to particular criteria through the choice of competing word sequences and the choice of smoothing. Analytic and experimental comparisons are presented for both the maximum mutual information (MMI) and the minimum classi®cation error (MCE) criterion together with the optimization methods gradient descent (GD) and extended Baum (EB) algorithm. A tree search-based restricted recognition method using word graphs is presented, so as to reduce the computational complexity of large vocabulary discriminative training. Moreover, for MCE training, a method using word graphs for ecient calculation of discriminative statistics is introduced. Experiments were performed for continuous speech recognition using the ARPA wall street journal (WSJ) corpus with a vocabulary of 5k words and for the recognition of continuously spoken digit strings using both the TI digit string corpus for American English digits, and the SieTill corpus for telephone line recorded German digits. For the MMI criterion, neither analytical nor experimental results do indicate signi®cant dierences between EB and GD optimization. For acoustic models of low complexity, MCE training gave signi®cantly better results than MMI training. The recognition results for large vocabulary MMI training on the WSJ corpus show a signi®cant dependence on the context length of the language model used for training. Best results were obtained using a unigram language model for MMI training. No signi®cant correlation has been observed between the language models chosen for training and recognition. Ó 2001 Elsevier Science B.V. All rights reserved. ZusammenfassungZiel dieser Arbeit ist die Schaung eines einheitlichen Rahmens f ur eine Klasse von diskriminativen Trainingskriterien und Optimierungsmethoden f ur die kontinuierliche Spracherkennung. Dazu wird ein einheitliches Kriterium de®niert, das auf Wahrscheinlichkeitsverh altnissen von korrekten und konkurrierenden Modellen basiert. Spezielle Kriterien ergeben sich daraus durch die Wahl der konkurrierenden Wortfolgen sowie der Gl attung. F ur die Kriterien maximum mutual information (MMI) und minimum classi®cation error (MCE), sowie deren Optimierung mittels Gradientenabstieg (GD) und erweitertem Baum (EB) Algorithmus werden analytische und experimentelle Vergleiche durchgef uhrt. Die Zeitkomplexit at des diskriminativen Trainings bei groûem Vokabular wurde durch eine Methode zur Einschr ankung der Baumsuche reduziert. R esum eLe but de ce travail est de d e®nir un cadre commun incluant un ensemble de crit eres d'apprentissage discriminant et de m ethodes d'optimisation pour la reconnaissance de la parole continue. Nous introduisons un crit ere discriminant fond e sur le rapport entre la vraissemblanc...

show abstract

Section: Discriminative Training Using Word Graphsmentioning

confidence: 99%

Comparison of discriminative training criteria

Schlüter

Macherey²

Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '98 (Cat. No.98CH36181

View full text Add to dashboard Cite

show abstract

“…A fast time-synchronous backward pass produces an inital N-best list using the Word-Dependent N-best algorithm [5].…”

Section: Byblosmentioning

confidence: 99%

Comparative experiments on large vocabulary speech recognition

Kubala

Anastasakos²,

Makhoul³

et al.

Proceedings of ICASSP '94. IEEE International Conference on Acoustics, Speech and Signal Processing

Self Cite

View full text Add to dashboard Cite

This paper describes several key experiments in large vocabulay speech recognition. We demonstrate that, counter to our intuitions, given a fixed amount of training speech, the number of training speakers has little effect on the accuracy. We show how much speech is needed for speaker-independent (SI) recognition in order to achieve the same performance as speakerdependent (SD) recognition. We demonstrate that, though the N-Best Paradigm works quite well up to vocabularies of 5,000 words, it begins to break down with 20,000 words and long sentences. We compare the performance of two feature preprocessing algorithms for microphone independence and we describe a new microphone adaptation algorithm based on selection among several codebook transformations.

show abstract

“…Lattice [7,144] and N -best [153] rescoring, as discussed in Architecture of an HMM-Based Recogniser, are standard approaches to speeding up decoding by reducing the search-space. Though normally described as approaches to improve efficiency, they have a side-effect of implicit system combination.…”

Section: Implicit Combinationmentioning

confidence: 99%

The Application of Hidden Markov Models in Speech Recognition

Gales

Young

2007

FNT in Signal Processing

427

155

View full text Add to dashboard Cite

Hidden Markov Models (HMMs) provide a simple and effective framework for modelling time-varying spectral vector sequences. As a consequence, almost all present day large vocabulary continuous speech recognition (LVCSR) systems are based on HMMs.Whereas the basic principles underlying HMM-based LVCSR are rather straightforward, the approximations and simplifying assumptions involved in a direct implementation of these principles would result in a system which has poor accuracy and unacceptable sensitivity to changes in operating environment. Thus, the practical application of HMMs in modern systems involves considerable sophistication.The aim of this review is first to present the core architecture of a HMM-based LVCSR system and then describe the various refinements which are needed to achieve state-of-the-art performance. These refinements include feature projection, improved covariance modelling, discriminative parameter estimation, adaptation and normalisation, noise compensation and multi-pass system combination. The review concludes with a case study of LVCSR for Broadcast News and Conversation transcription in order to illustrate the techniques described.

show abstract

A comparison of several approximate algorithms for finding multiple (N-best) sentence hypotheses

Cited by 110 publications

References 8 publications

Comparison of discriminative training criteria

Comparison of discriminative training criteria

Comparative experiments on large vocabulary speech recognition

The Application of Hidden Markov Models in Speech Recognition

Contact Info

Product

Resources

About