Utterance verification of keyword strings using word-based minimum verification error (WB-MVE) training

Sukkar, R.A.; Setlur, A.R.; Rahim, Mazin G.; Lee, Chin‐Hui

doi:10.1109/icassp.1996.541147

Cited by 35 publications

(26 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In other words, these approaches do not propose to train the model so as to maximize the spotting performance, and the keyword spotting task is only introduced in the inference step after training. Only few studies have proposed discriminative parameter training approaches to circumvent this weakness (Benayed et al 2003;Sandness and Hetherington 2000;Sukkar et al 1996;Weintraub et al 1997). Sukkar et al (1996) proposed to maximize the likelihood ratio between the keyword and garbage models for keyword utterances and to minimize it over a set of false alarms generated by a first keyword spotter.…”

Section: Previous Workmentioning

confidence: 99%

“…Only few studies have proposed discriminative parameter training approaches to circumvent this weakness (Benayed et al 2003;Sandness and Hetherington 2000;Sukkar et al 1996;Weintraub et al 1997). Sukkar et al (1996) proposed to maximize the likelihood ratio between the keyword and garbage models for keyword utterances and to minimize it over a set of false alarms generated by a first keyword spotter. Sandness and Hetherington (2000) proposed to apply Minimum Classification Error (MCE) to the keyword spotting problem.…”

Section: Previous Workmentioning

confidence: 99%

See 1 more Smart Citation

Discriminative keyword spotting

Keshet

Grangier²,

Bengio

2009

Speech Communication

View full text Add to dashboard Cite

This chapter introduces a discriminative method for detecting and spotting keywords in spoken utterances. Given a word represented as a sequence of phonemes and a spoken utterance, the keyword spotter predicts the best time span of the phoneme sequence in the spoken utterance along with a confidence. If the prediction confidence is above certain level the keyword is declared to be spoken in the utterance within the predicted time span, otherwise the keyword is declared as not spoken. The problem of keyword spotting training is formulated as a discriminative task where the model parameters are chosen so the utterance in which the keyword is spoken would have higher confidence than any other spoken utterance in which the keyword is not spoken. It is shown theoretically and empirically that the proposed training method resulted with a high area under the receiver operating characteristic (ROC) curve, the most common measure to evaluate keyword spotters. We present an iterative algorithm to train the keyword spotter efficiently. The proposed approach contrasts with standard spotting strategies based on HMMs, for which the training procedure does not maximize a loss directly related to the spotting performance. Several experiments performed on TIMIT and WSJ corpora show the advantage of our approach over HMM-based alternatives.

show abstract

Section: Previous Workmentioning

confidence: 99%

Section: Previous Workmentioning

confidence: 99%

Discriminative keyword spotting

Keshet

Grangier²,

Bengio

2009

Speech Communication

View full text Add to dashboard Cite

show abstract

“…Typically, likelihood ratio testing (LRT)-based confidence measures are formulated for this purpose, and it was discovered that some discriminative training techniques, such as minimum classification error (MCE) and minimum verification error (MVE) [3] methods, can significantly improve the performance of utterance verification. These LRT-based confidence measures have shown good performance with posterior probability-based confidence measures [1], [4].…”

Section: Introductionmentioning

confidence: 99%

Utterance Verification Using Word Voiceprint Models Based on Probabilistic Distributions of Phone-Level Log-Likelihood Ratio and Phone Duration

Kwon¹,

Kim²

2008

IEICE Transactions on Information and Systems

View full text Add to dashboard Cite

SUMMARY This paper suggests word voiceprint models to verify the recognition results obtained from a speech recognition system. Word voiceprint models have word-dependent information based on the distributions of phone-level log-likelihood ratio and duration. Thus, we can obtain a more reliable confidence score for a recognized word by using its word voiceprint models that represent the more proper characteristics of utterance verification for the word. Additionally, when obtaining a loglikelihood ratio-based word voiceprint score, this paper proposes a new log-scale normalization function using the distribution of the phone-level log-likelihood ratio, instead of the sigmoid function widely used in obtaining a phone-level log-likelihood ratio. This function plays a role of emphasizing a mis-recognized phone in a word. This individual information of a word is used to help achieve a more discriminative score against out-ofvocabulary words. The proposed method requires additional memory, but it shows that the relative reduction in equal error rate is 16.9% compared to the baseline system using simple phone log-likelihood ratios.

show abstract

“…The problem is formulated at the word-level and subsequently extended to the string-level. It comprises a likelihood ratio test that is a function of two models (null and alternate hypothesis) which are both trained using a discriminative minimum verification error training (MVE) framework [1]. The null hypothesis is that the recognized word is correct while the alternate hypothesis is that the word is misrecognized.…”

Section: Mve-trained Likelihood Ratiomentioning

confidence: 99%

“…On the other hand, in applications such as telephone number dialing, it may be useless to get a string correct in a subsequent hypothesis if the top hypothesis is wrong. Utterance verification techniques have been successfully employed to identify and reject strings that are likely to be in error, thereby improving overall postrejection accuracy [1][2][3][4].…”

Section: Introductionmentioning

confidence: 99%

Correcting recognition errors via discriminative utterance verification

Setlur¹,

Sukkar²,

Jacob³

Proceeding of Fourth International Conference on Spoken Language Processing. ICSLP '96

Self Cite

View full text Add to dashboard Cite

Utterance verification (UV) is a process by which the output of a speech recognizer is verified to determine if the input speech actually includes the recognized keyword(s). The output of the speech verifier is a binary decision to accept or reject the recognized utterance based on a UV confidence score. In this paper, we extend the notion of utterance verification to not only detect errors but also selectively correct them. We perform error correction by flipping the hypotheses produced by an N-best recognizer in cases when the top candidate has a UV confidence score that is lower than that of the next candidate. We propose two measures for computing confidence scores and investigate the use of a hybrid confidence measure that combines the two measures into a single score. Using this hybrid confidence measure and an N-best algorithm, we obtained an 11% improvement in word-error rate on a connected digit recognition task. This improvement was achieved while still maintaining reliable detection of nonkeyword speech and misrecognitions.

show abstract

Utterance verification of keyword strings using word-based minimum verification error (WB-MVE) training

Cited by 35 publications

References 6 publications

Discriminative keyword spotting

Discriminative keyword spotting

Utterance Verification Using Word Voiceprint Models Based on Probabilistic Distributions of Phone-Level Log-Likelihood Ratio and Phone Duration

Correcting recognition errors via discriminative utterance verification

Contact Info

Product

Resources

About