CoChat
Sign Up
Cited by 119Speech Recognition and SynthesisSpeech and dialogue systemsSpeech and Audio Processing

Discriminative utterance verification for connected digits recognition

Rahim, M.G., Chin-Hui Lee, Biing-Hwang Juang·1997·IEEE Transactions on Speech and Audio Processing
DOI 10.1109/89.568733
Open this paper in CoChat — free account

Abstract

Utterance verification represents an important technology in the design of user-friendly speech recognition systems. It involves the recognition of keyword strings and the rejection of nonkeyword strings. This paper describes a hidden Markov model-based (HMM-based) utterance verification system using the framework of statistical hypothesis testing. The two major issues on how to design keyword and string scoring criteria are addressed. For keyword verification, different alternative hypotheses are proposed based on the scores of antikeyword models and a general acoustic filler model. For string verification, different measures are proposed with the objective of detecting nonvocabulary word strings and possibly erroneous strings (so-called putative errors). This paper also motivates the need for discriminative hypothesis testing in verification. One such approach based on minimum classification error training is investigated in detail. When the proposed verification technique was integrated into a state-of-the-art connected digit recognition system, the string error rate for valid digit strings was found to decrease by 57% when setting the rejection rate to 5%. Furthermore, the system was able to correctly reject over 99.9% of nonvocabulary word strings.

Integrity signals

Registry

Crossref · 10.1109/89.568733

Cites

2 of 2 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • If the probabilistic assumptions in an HMM or other probabilistic model are not realistic, discriminative training can improve performance, as shown for speech recognition systems.

    “On the other hand, if the probabilistic assumptions in an HMM (or other probabilistic model) are not realistic, discriminative training, discussed in Section VI, can improve performance as this has been clearly shown for speech recognition systems [48]- [50], [107], [108].”

    From Gradient-based learning applied to document recognition (1998) · cites this paper as [108]

Cites

Flagged first, then by citations

Research works printed but not linked: 35

  1. 10.1109/ICASSP.1993.319338 · DOI
  2. 10.1109/ICASSP.1994.389273 · DOI
  3. 10.1109/ICASSP.1994.389331 · DOI
  4. rosenberg, 1992, the use of cohort normalized scores for speaker recognition, Proc Int Conf Spoken Language Processing, 599
  5. 10.1109/29.103088 · DOI
  6. 10.1109/ICASSP.1993.319341 · DOI
  7. 10.1109/ICASSP.1993.319339 · DOI
  8. 10.1109/ICASSP.1996.541147 · DOI
  9. fukunaga, 1990, Statistical Pattern Recognition
  10. 10.1109/78.175747 · DOI
  11. 10.1109/NNSP.1991.239540 · DOI
  12. lamel, 1986, speech database development: design and analysis of the acoustic-phonetic corpus, Proc Speech Recognition Workshop (DARPA)
  13. 10.1121/1.412286 · DOI
  14. moreno, 1990, rejection techniques in continuous speech recognition using hidden markov models, Proc Europ Conf Signal Processing, 1383
  15. rabiner, 1993, Fundamentals of speech recognition
  16. rabiner, 1978, Digital Processing of Speech Signals
  17. 10.1109/ICASSP.1995.479529 · DOI
  18. chang, 1993, figure of merit training for detection and spotting, Proc Conf Neural Information Processing Systems, 6, 1019
  19. rose, 1991, techniques for robust wordspotting in continuous speech messages, Proc Europ Conf Speech Communications, 1183
  20. bourlard, 1994, optimizing recognition and rejection performance in wordspotting systems, Proc Int Conf on Acoustics Speech and Signal Processing, i, 373
  21. 10.1109/ICASSP.1990.115555 · DOI
  22. 10.1109/ICASSP.1992.225869 · DOI
  23. 10.1109/ICASSP.1992.226112 · DOI
  24. 10.1109/ICASSP.1993.319342 · DOI
  25. chou, 1994, a high resolution <formula><tex>$n$</tex></formula>-best search algorithm using inter-word context dependent models for continuous speech recognition, Proc Int Conf on Acoustics Speech and Signal Processing
  26. 10.1142/S0218001494000024 · DOI
  27. bickel, 1977, Mathematical Statistics Basic Ideas and Selected Topics
  28. feng, 1992, continuous word spotting for applications in telecommunications, Proc Int Conf on Spoken Language Processing, 21
  29. 10.1109/97.489062 · DOI
  30. 10.1109/ICASSP.1996.550804 · DOI
  31. 10.1109/TSA.1996.481449 · DOI
  32. 10.1109/ICASSP.1989.266505 · DOI
  33. 10.1121/1.418519 · DOI
  34. rose, 1992, discriminant wordspotting techniques for rejecting nonvocabulary utterances in unconstrained speech, Proc Int Conf Acoustics Speech and Signal Processing, ii, 105
  35. 10.1109/ICASSP.1993.319340 · DOI

Cited by

Sorted by citations

Authors
Rahim, M. G., Lee, C.-H., & Juang, B.-H.
Affiliations
  • AT&T (United States) (US)
  • Alcatel-Lucent (France) (FR)
ORCID
Venue
IEEE Transactions on Speech and Audio Processing
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Published
1997-05
Volume
5
Issue
3
Pages
266-277
ISSN
1063-6676 (ISSN-L)
DOI
10.1109/89.568733
OpenAlex
W2135936685
Cited by
119
References
2
Citations by year
2024: 1 · 2021: 1 · 2020: 3 · 2019: 2 · 2018: 3 · 2016: 3 · 2015: 4 · 2014: 1 · 2013: 6 · 2012: 4
FWCI
7.48
Type
journal article
Language
en
Open access
closed
Licence
ieeexplore.ieee.org/Xplorehelp/downloads/license-information/IEEE.html (vor)
Locations
IEEE Transactions on Speech and Audio Processing (publishedVersion)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Speech Recognition and Synthesis
  • Physical Sciences › Computer Science › Artificial Intelligence › Speech and dialogue systems
  • Physical Sciences › Computer Science › Signal Processing › Speech and Audio Processing
Keywords
utterance verification, connected digit recognition, keyword spotting, hidden Markov models, hypothesis testing, minimum classification error training, discriminative training

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review