Hong Liu
ABSTRACT
Keyword1 spotting (KWS) deals with the identification of keywords in speech utterances. A two-stage approach is often used for the flexibility and high efficiency. The two stages are keyword hypotheses detection stage and hit or false-alarm verification stage in sequence. How to reduce the false-alarms is a key and difficult problem in the verification stage, which is formatted as the confidence measure (CM) problem. In this paper, a novel keyword-filler hidden Markov model (HMM) based method is proposed based on two improved approaches. On one hand, for more effective confidence measure, a hypothesis boundary realignment method is used to gain more precise hypothesized segments for possible keyword. Then an overlap ratio criterion is defined to evaluate this process. On the other hand, a state-level confidence weighting method is proposed to improve the posterior probability based CM. Experiments show that either improvement is effective, and the proposed method based on the two processes gives the best performance.
- R. C. Rose and D. B. Paul. 1990. A hidden Markov model based keyword recognition system. In International Conference on Acoustics, Speech and Signal Processing, IEEE, 1, 129--132.Google Scholar
- J. R. Rohlicek, W. Russell, S. Roukos, and H. Gish. 1989. Continuous hidden Markov modeling for speaker-independent word spotting. In International Conference on Acoustics, Speech and Signal Processing, IEEE, 1, 627--630.Google Scholar
- R. L. Warren. 2001. Broadcast speech recognition system for keyword monitoring. U.S. Patent 6332120 B1.Google Scholar
- J. S. Garofolo, C. G. P. Auzanne, and E. M. Voorhees. 2000. The TREC spoken document retrieval track: a success story. In Text Retrieval Conference, NIST, 26, 1--20. Google ScholarDigital Library
- G. G. Chen, C. Parada, and G. Heigold. 2014. Small-footprint keyword spotting using deep neural networks. In International Conference on Acoustics, Speech and Signal Processing, IEEE, 4087--4091.Google Scholar
- A. H. Michaely, X. D. Zhang, G. Simko, C. Parada, and P. Aleksic. 2017. Keyword spotting for Google assistant using contextual speech recognition. In Automatic Speech Recognition and Understanding Workshop, IEEE, 272--278.Google Scholar
- D. R. H. Miller, M. Kleber, C. Kao, O. Kimball, T. Colthurst, S. A. Lowe, R. M. Schwartz, and H. Gish. 2007. Rapid and accurate spoken term detection. In International Conference of the Speech Communication Association, 314--317.Google Scholar
- J. Mamou, B. Ramabhadran, and O. Siohan. 2007. Vocabulary independent spoken term detection. In International ACM SIGIR Conference on Research and Development in Information Retrieval, 615--622. Google ScholarDigital Library
- D. Vergyri, I. Shafran, A. Stolcke, R. R. V. Gadde, M. Akbacak, B. Roark, and W. Wang. 2007. The SRI/OGI 2006 spoken term detection system. In International Conference of the Speech Communication Association, 2393--2396.Google Scholar
- V. T. Pham, H. H. Xu, X. Xiao, N. F. Chen, E. S. Chng, and H. Z. Li. 2016. Keyword search using query expansion for graph-based rescoring of hypothesized detections. In International Conference on Acoustics, Speech and Signal Processing, IEEE, 6035--6039.Google Scholar
- T. Alume, D. Karakos, W. Hartmann, R. Hsiao, L. Zhang, L. Nguyen, S. Tsakalidis, and R. Schwartz. 2017. The 2016 BBN Georgian telephone speech keyword spotting system. In International Conference on Acoustics, Speech and Signal Processing, IEEE, 5755--5759.Google Scholar
- D. Karakos and R. M. Schwartz. 2015. Combination of search techniques for improved spotting of OOV keywords. In International Conference on Acoustics, Speech and Signal Processing, IEEE, 5336--5340.Google Scholar
- I. Szöke. 2010. Hybrid word-subword spoken term detection.Google Scholar
- K. M. Knill and S. J. Young. 1996. Fast implementation methods for Viterbi-based word-spotting. In International Conference on Acoustics, Speech and Signal Processing, IEEE, 522--525. Google ScholarDigital Library
- M. Weintraub. 1995. LVCSR log-likelihood ratio scoring for keyword spotting. In International Conference on Acoustics, Speech and Signal Processing, IEEE, 1, 297--300.Google Scholar
- H. Bourlard, B. D'Hoore, and J. M. Boite. 1994. Optimizing recognition and rejection performance in wordspotting systems. In International Conference on Acoustics, Speech and Signal Processing, IEEE, 1, I/373--I/376.Google Scholar
- R. A. Sukkar and C. H. Lee. 1996. Vocabulary independent discriminative utterance verification for non-keyword rejection in subword based speech recognition. IEEE Transactions on Speech and Audio Processing, 4(6), 420--429.Google ScholarCross Ref
- S. Abdou and M. S. Scordilis. 2004. Beam search pruning in speech recognition using a posterior probability-based confidence measure. Speech Communication, 42(3), 409--428.Google ScholarCross Ref
- W. E. Fisher, G. R. Doddington, and K. M. Goudle-Marshall. 1986. The DARPA speech recognition research database: specifications and status. CMU Arctic Speech Databases for Speech Synthesis Research.Google Scholar
- S. Young, G. Evermann, M. Gales, T. Hain, D. Kershaw, X. Liu, G. Moore, J. Odell, D. Ollason, D. Povey, V. Valtchev, and P. Woodland. 2009. The HTK Book (for HTK version 3.4.1), http://htk.eng.cam.ac.uk: Cambridge University.Google Scholar
- J. Liang, M. Meng, X. R. Wang, and P. Ding. 2006. An improved Mandarin keyword spotting system using MCE training and context-enhanced verification. In International Conference on Acoustics, Speech and Signal Processing, IEEE, 1145--1148.Google Scholar
- H. Y. Li, J. Q. Han, and T. R. Zheng. 2011. AUC optimization based confidence measure for keyword spotting. In International Conference of the Speech Communication Association, 1917--1920.Google Scholar
- Y. C. Liu, M. X. Xu, and L. H. Cai. 2014. Improved keyword spotting system by optimizing posterior confidence measure vector using feed-forward neural network. In International Joint Conference on Neural Networks, IEEE, 2036--2041.Google Scholar
- C. Cortes and M. Mohri. 2004. Confidence intervals for the area under the ROC curve. In International Conference on Neural Information Processing Systems, MIT Press, 305--312. Google ScholarDigital Library
- K. F. Lee and H. W. Hon. 1989. Speaker-independent phone recognition using hidden Markov models. IEEE Transactions on Acoustics Speech and Signal Processing, 37(11), 1641--1648.Google ScholarCross Ref
- J. G. Fiscus, J. G. Ajot, J. Garofalo, and G. Doddington. 2007. Results of the 2006 spoken term detection evaluation. In SIGIR Workshop on Searching Spontaneous Conversational Speech, 51--57.Google Scholar
- J. Cui, X. D. Cui, B. Ramabhadran, J. Kim, B. Kingsbury, J. Mamou, L. Mangu, M. Picheny, T. N. Sainath, and A. Sethy. 2013. Developing speech recognition systems for corpus indexing under the IARPA Babel program. In International Conference on Acoustics, Speech and Signal Processing, IEEE, 6753--6757.Google Scholar
Index Terms
- Keyword Spotting Based on Hypothesis Boundary Realignment and State-Level Confidence Weighting
Recommendations
Word spotting based ona posterior measure of keyword confidence
AbstractIn this paper, an approach of keyword confidence estimation is developed that well combines acoustic layer scores and syllable-based statistical language model (LM) scores. Ana posteriori (AP) confidence measure and its forward-backward ...
A novel keyword rescoring method for improved spoken keyword spotting
AbstractIn this paper, we present a spoken KeyWord Spotting (KWS) system which creates a search index from word lattices generated by a deep speech recognizer. Basic KWS systems estimate word posteriors from the lattices and use them to make “correct/...
Comparison of Confidence Measures for Face Recognition
FG '00: Proceedings of the Fourth IEEE International Conference on Automatic Face and Gesture Recognition 2000This paper compares different confidence measures for the results of statistical face recognition systems. The main applications of a confidence measure are rejection of unknown people and the detection of recognition errors. Some of the confidence ...
Comments