Joint decoding of tandem and hybrid systems for improved keyword spotting on low resource languages

Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH

Volumn 2015-January, Issue , 2015, Pages 3660-3664

  Wang, Haipeng ^a   Ragni, Anton ^a   Gales, Mark J F ^a   Knill, Kate M ^a   Woodland, Philip C ^a   Zhang, Chao ^a  

^a UNIVERSITY OF CAMBRIDGE   (United Kingdom)

Author keywords

Deep neural network; Hybrid; Joint decoding; Keyword spotting; Tandem

Indexed keywords

COMPUTATIONAL LINGUISTICS; DECODING; HYBRID SYSTEMS; SEARCH ENGINES; SPEECH COMMUNICATION;

DEEP NEURAL NETWORKS; HYBRID; JOINT DECODING; KEYWORD SPOTTING; TANDEM;

SPEECH RECOGNITION;

EID: 84959166110     PISSN: 2308457X     EISSN: 19909772     Source Type: Conference Proceeding    
DOI: None     Document Type: Conference Paper

References (43)

1. L. Mangu, H. Soltau, H.-K. Kuo, B. Kingsbury, and G. Saon, "Exploitingdiversity for spoken term detection, " in Proc. ICASSP, 2013, pp. 8282-8286.
2. D. Karakos, R. Schwartz, S. Tsakalidis, L. Zhang, S. Ranjan, T. Ng, and R. Hsiao, "Score normalization and system combinationfor improved keyword spotting, " in Proc. ASRU, 2013, pp. 210-215.
3. A. Ragni, K. Knill, S. Rath, and M. Gales, "Data augmentationfor low resource languages, " in Proc. Interspeech, 2014, pp. 810-814.
4. K. Knill, M. Gales, A. Ragni, and S. Rath, "Language independentand unsupervised acoustic models for speech recognition and keyword spotting, " in Proc. INTERSPEECH, 2014, pp. 20-26.
5. M. Harper, "IARPA Solicitation IARPA-BAA-11-02, " 2011, http: //www. iarpa. gov/solicitations babel. html.
6. J. Lee, "Analyses of multiple evidence combination, " in ACM SIGIR, 1997, pp. 267-276.
7. J. Mamou, J. Cui, X. Cui, M. Gales, B. Kingsbury, K. Knill, L. Mangu, D. Nolden, M. Picheny, B. Ramabhadran et al., "Systemcombination and score normalization for spoken term detection, "in Proc. ICASSP, 2013, pp. 8272-8276.
8. N. Chen et al., "Low-resource keyword search strategies forTAMIL, " in Proc. ICASSP, 2015, pp. 5366-5370.
9. J. G. Fiscus, "A Post-processing System to Yield ReducedWord Error Rates: Recogniser Output Voting Error Reduction(ROVER), " in Proc. ASRU, 1997, pp. 347-354.
10. G. Evermann and P. Woodland, "Posterior Probability Decoding, Confidence Estimation and System Combination, " in Proc. Speech Transcription Workshop, vol. 27, 2000.
11. D. Miller, M. Kleber, C.-L. Kao, O. Kimball, T. Colthurst, S. Lowe, R. Schwartz, and H. Gish, "Rapid and accurate spokenterm detection, " in Proc. Interspeech, 2007.
12. D. Vergyri, I. Shafran, A. Stolcke, V. Gadde, M. Akbacak et al., "The SRI/OGI 2006 spoken term detection system, " in Proc. Interspeech, 2007, pp. 2393-2396.
13. I. Szoke, L. Burget, J. Cernocky, and M. Fapso, "Sub-word modelingof out of vocabulary words in spoken term detection, " Proc. SLT, 2008, pp. 273-276.
14. B. Kingsbury et al., "A high-performance Cantonese keywordsearch system, " in Proc. ICASSP, 2013, pp. 8277-8281.
15. S. Rath, K. Knill, A. Ragni, and M. Gales, "Combining tand emand hybrid systems for improved speech recognition and keywordspotting on low resource languages, " in Proc. Interspeech, 2014, pp. 835-839.
16. J. Park, F. Diehl, M. Gales, M. Tomalin, and P. C. Woodland, "Theefficient incorporation of MLP features into automatic speechrecognition systems, " Computer Speech and Language, vol. 25, no. 3, pp. 519-534, 2010.
17. A. Mohamed, G. Dahl, and G. Hinton, "Acoustic modeling usingdeep belief networks, " Audio, Speech, and Language Processing, IEEE Transactions on, vol. 20, no. 1, pp. 14-22, 2012.
18. G. Hinton, L. Deng, D. Yu, G. Dahl, A. Mohamed, N. Jaitly, A. Senior, V. Vanhoucke, P. Nguyen, T. Sainath et al., "Deep neuralnetworks for acoustic modeling in speech recognition, " IEEESignal Processing Magazine, vol. 29, no. 6, pp. 82-97, Nov 2012.
19. H. Bourlard and N. Morgan, Connectionist Speech Recognition: A Hybrid Approach. Springer Science & Business Media, 1994, vol. 247.
20. A. Morris, A. Hagen, H. Glotin, and H. Bourlard, "Multi-streamadaptive evidence combination for noise robust ASR, " SpeechCommunication, vol. 34, no. 1, pp. 25-40, 2001.
21. H. Misra, H. Bourlard, and V. Tyagi, "New entropy based combinationrules in HMM/ANN multi-stream ASR, " in Proc. ICASSP, 2003, pp. 738-741.
22. S. Renals, N. Morgan, H. Bourlard, M. Cohen, and H. Franco, "Connectionist probability estimators in hmm speech recognition, "IEEE Trans. Speech and Audio Processing, vol. 2, no. 1, pp. 161-174, 1994.
23. C. Dugast, L. Devillers, and X. Aubert, "Combining TDNN and HMM in a hybrid system for improved continuous-speech recognition, "IEEE Trans. Speech and Audio Processing, vol. 2, no. 1, pp. 217-223, 1994.
24. P. Swietojanski, A. Ghoshal, and S. Renals, "Revisiting hybridand gmm-hmm system combination techniques, " in Proc. ICASSP, 2013, pp. 6744-6748.
25. X. He and K. Toutanova, "Joint optimization for machine translationsystem combination, " in Proc. EMNLP, 2009, pp. 1202-1211.
26. H. Soltau, G. Saon, and T. Sainath, "Joint training of convolutionaland non-convolutional neural networks, " Proc. ICASSP, 2014.
27. J. Fiscus, J. Ajot, J. Garofolo, and G. Doddingtion, "Results of the2006 Spoken Term Detection Evaluation, " in Proc. SIGIR, 2007, pp. 51-57.
28. S. Young, G. Evermann, M. Gales, T. Hain, D. Kershaw, X. Liu, G. Moore, J. Odell, D. Ollason, D. Povey et al., The HTK Book(for HTK version 3. 4. 1). http: //htk. eng. cam. ac. uk: CambridgeUniversity, 2009.
29. C. Zhang and P. Woodland, "A general artificial neural networkextension for HTK, " in Submission to InterSpeech, 2015.
30. M. Gales, K. Knill, and A. Ragni, "Unicode-based graphemic systemsfor limited resource languages, " in Proc. ICASSP, 2015.
31. L. Lamel and J.-L. Gauvain, "Lightly supervised and unsupervisedacoustic model training, " Computer speech and language, vol. 16, pp. 115-129, 2013.
32. S. Thomas, M. L. Seltzer, K. Church, and H. Hermansky, "Deepneural network features and semi-supervised training for low resourcespeech recognition, " in Proc. ICASSP, 2013, pp. 6704-6708.
33. R. Hsiao, T. Ng, F. Grézl, D. Karakos, S. Tsakalidis, L. Nguyen, and R. Schwartz, "Discriminative semi-supervised training forkeyword search in low resource languages, " in Proc. ASRU, 2013, pp. 440-445.
34. Z. Tuske, J. Pinto, D. Willett, and R. Schluter, "Investigation oncross-and multilingual MLP features under matched and mismatchedacoustical conditions, " in Proc. ICASSP, 2013, pp. 6970-6974.
35. Z. Tuske, D. Nolden, R. Schluter, and H. Ney, "MultilingualMRASTA features for low-resource keyword search and speechrecognition systems, " in Proc. ICASSP, 2014, pp. 7854-7858.
36. D. Povey et al., "The Kaldi speech recognition toolkit, " in Proc. ASRU, 2011.
37. M. Gales, "Semi-tied covariance matrices for hidden markovmodels, " Speech and Audio Processing, IEEE Transactions on, vol. 7, no. 3, pp. 272-281, 1999.
38. G. Saon, S. Dharanipragada, and D. Povey, "Feature space gaussianization, "in Proc. ICASSP, 2004, p. 326329.
39. D. Povey and P. C. Woodland, "Minimum Phone Error and I-smoothing for improved discriminative training, " in Proc. ICASSP, 2002, pp. 101-105.
40. T. Anastasakos, J. McDonough, R. Schwartz, and J. Makhoul, "Acompact model for speaker adaptive training, " in Proc. ICSLP, 1996, pp. 1137-1140.
41. M. Gales, "Maximum likelihood linear transformations forHMM-based speech recognition, " Computer speech & language, vol. 12, no. 2, pp. 75-98, 1998.
42. K. Vesely, A. Ghoshal, L. Burget, and D. Povey, "Sequencediscriminativetraining of deep neural networks. " in Proc. Interspeech, 2013, pp. 2345-2349.
43. M. Mohri, C. Allauzen, and M. Saraclar, "General indexation ofweighted automata-application to spoken utterance retrieval, " Proc. HLT/NAACL, 2004, pp. 33-40.