Integration of acoustic information in a large vocabulary word recognizer

Gupta, Vishwa; Lennig, M.; Mermelstein, P.

doi:10.1109/icassp.1987.1169578

Cited by 61 publications

(20 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Also, in his experiments, each word in the 100(X)-WOK1 dictionary has been uttered and is tested once, whereas our experiments deal with words in sentences, where frequent words (generally the shortest and the most ambiguous) occur several times. The same considerations apply to the work of Gupta et al, reported in [7].…”

Section: Recognition Experimentssupporting

confidence: 60%

See 1 more Smart Citation

Multilevel decoding for Very-Large-Size-Dictionary speech recognition

Mérialdo

1988

IBM J. Res. & Dev.

View full text Add to dashboard Cite

An important concern in the field of speech recognition is the size of the vocabulary that a recognition system is able to support. Large vocabularies introduce difficulties involving the amount of computation the system must perform and the number of ambiguities it must resolve. But, for practical applications in general and for dictation tasks in particular, large vocabularies are required, because of the difficulties and inconveniences involved in restricting the speaker to the use of a limited vocabulary. This paper describes a new organization of the recognition process, Multilevel Decoding (MLD), that allows the system to support a Very-LargeSize Dictionary (VLSD)-one comprising over 100000 words. This significantly surpasses the capacity of previous speech-recognition systems. With MLD, the effect of dictionary size on the accuracy of recognition can be studied. In this paper, recognition experiments using 10000-and 200000-word dictionaries are compared. They indicate that recognition using ^Copyright 1988 by International Business Machines Corporation. Copying in printed form for private use is permitted without payment of royalty provided that (1) each reproduction is done without alteration and (2) the Journal reference and IBM copyright notice are included on the first page. The title and abstract, but no other portions, of this paper may be copied or distributed royalty free without further permission by computer-based and other information-service systems. Permission to republish any other portion of this paper must be obtained from the Editor. a 200000-word dictionary is more accurate than recognition using a 10000-word dictionary (when unrecognized words are included in the error rate).

show abstract

Section: Recognition Experimentssupporting

confidence: 60%

“…Recognition accuracy was 94% when using word templates and 88% when using syllable templates. Gupta et al [7] have studied a special class of Markov models to recognize items from a 60000-homophone-set English dictionary, spoken in isolated mode. Recognition accuracy varied from 52% to 76%, depending on the choice of model.…”

Section: Introductionmentioning

confidence: 99%

Multilevel decoding for Very-Large-Size-Dictionary speech recognition

Mérialdo

1988

IBM J. Res. & Dev.

View full text Add to dashboard Cite

show abstract

“…Second, when g t ( i )= i and φ t ( i,k )=b i (the regression coefficients are not explicitly dependent on the time index within a state), the model would become the hidden filter model or linear-predictive HMM (Poritz, 1988;Kenny et al, 1990). Third, if g t ( i )= i and φ t ( i,k )=+1, the model would be degenerated to the dynamic-parameter HMM (Gupta, Lennig & Mermelstein, 1987). Finally, removing the autoregression terms (K= 0), while leaving all other terms in the most general form, would revert the model to the first-order non-stationary-state HMM developed previously by Deng (1992).…”

Section: Discussionmentioning

confidence: 97%

A Markov model containing state-conditioned second-order non-stationarity: application to speech recognition

Deng

Rathinavelu

1995

Computer Speech & Language

View full text Add to dashboard Cite

“…This formulation proposed by Gupta et. al., [11] has been used in highly successful Sphinx-1 system [3].…”

Section: Literature Surveymentioning

confidence: 99%

Offline handwritten word recognition using MQDF-HMMs

et al. 2015

View full text Add to dashboard Cite

We propose an improved HMM formulation for offline handwriting recognition (HWR). The main contribution of this work is using modified quadratic discriminant function (MQDF) [1] within HMM framework. In an MQDF-HMM the state observation likelihood is calculated by a weighted combination of MQDF likelihoods of individual Gaussians of GMM (Gaussian Mixture Model). The quadratic discriminant function (QDF) of a multivariate Gaussian can be rewritten by avoiding the inverse of covariance matrix by using the Eigen values and Eigen vectors of it. The MQDF is derived from QDF by substituting few of badly estimated lower-most Eigen values by an appropriate constant. The estimation errors of non-dominant Eigen vectors and Eigen values of covariance matrix for which the training data is insufficient can be controlled by this approach. MQDF has been successfully shown to improve the character recognition performance [1]. The usage of MQDF in HMM improves the computation, storage and modeling power of HMM when there is limited training data. We have got encouraging results on offline handwritten character (NIST database) and word recognition in English using MQDF HMMs.

show abstract

Integration of acoustic information in a large vocabulary word recognizer

Cited by 61 publications

References 8 publications

Multilevel decoding for Very-Large-Size-Dictionary speech recognition

Multilevel decoding for Very-Large-Size-Dictionary speech recognition

A Markov model containing state-conditioned second-order non-stationarity: application to speech recognition

Offline handwritten word recognition using MQDF-HMMs

Contact Info

Product

Resources

About