New n-best based rejection techniques for improving a real-time telephonic connected word recognition system

Caminero-Gil, F. Javier; Torre-Munilla, Celinda de la; Gómez, Luis A. Hernández; Alamo, Cesar Martín del

doi:10.21437/eurospeech.1995-502

Cited by 9 publications

(2 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In Table 1 we show the results of applying on-line garbage model to this task, together with the previous results obtained by using an explicit garbage model, corresponding to an improved baseline system of the one presented in [10]. We use two different rejection levels, to compare both systems, and a 3-candidates N-best recognition system.…”

Section: Resultsmentioning

confidence: 99%

On-line garbage modeling with discriminant analysis for utterance verification

Caminero-Gil¹,

Torre²,

Villarrubia³

et al. 1996

4th International Conference on Spoken Language Processing (ICSLP 1996)

View full text Add to dashboard Cite

Out-of-vocabulary (OOV) utterance detection and rejection are specially important and difficult problems in large-vocabulary and continuous speech recognition. In [1] we proposed an utterance verification procedure based on the use of frame-by-frame best acoustic state scores instead of using explicit garbage models. This procedure is usually referred to as on-line garbage modeling.In this contribution we extend our previous work in two major directions: a) we analyze, through the use of Discriminant Analysis, the possibilities of using L-best local scores and N-best utterance hypotheses scores for utterance verification; b) we present experimental results not only for a spontaneously spoken natural number recognition task, as in [1], but also for a flexible large vocabulary recognition task. All the results, based on a telephone database, show that the proposed on-line garbage modeling procedure outperforms, both in performance and computational cost, to other approaches based on the use of explicit garbage models.

show abstract

Section: Resultsmentioning

confidence: 99%

On-line garbage modeling with discriminant analysis for utterance verification

Caminero-Gil¹,

Torre²,

Villarrubia³

et al. 1996

4th International Conference on Spoken Language Processing (ICSLP 1996)

View full text Add to dashboard Cite

show abstract

“…Another measure for hypothesis testing that has been used widely, is the duration normalized likelihood ratio or log likelihood difference between the recognition hypothesis and the next best hypothesis [5,6]. The confidence score for hypothesis k is the likelihood ratio of candidates k and k+1 and is represented by (6) where h k and h k + 1 represent the k th and k + 1 th hypotheses produced by an N-best algorithm and O is the observation vector sequence. Unlike the MVE method, this approach requires no additional models to perform utterance verification.…”

Section: N-best Based Likelihood Ratiomentioning

confidence: 99%

Correcting recognition errors via discriminative utterance verification

Setlur¹,

Sukkar²,

Jacob³

1996

4th International Conference on Spoken Language Processing (ICSLP 1996)

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