Progress in dynamic programming search for LVCSR

Proceedings of the IEEE

Volumn 88, Issue 8, 2000, Pages 1224-1240

  Hermann Ney, Tefan Ortmanns ^a,b  

^a RWTH AACHEN UNIVERSITY   (Germany)

^b Svstaix GmbH   (Germany)

Author keywords

Bayes' decision rule; Beam search; Continuous speech recognition; Dynamic programming; Language model look ahead; Large vocabulary recognition; Lexical prefix tree organization; Pruning; Search organization; Word graph

Indexed keywords

EID: 0000635720     PISSN: 00189219     EISSN: None     Source Type: Journal    
DOI: 10.1109/5.880081     Document Type: Article

References (87)

1. F. Alleva, X. Huang, and M.-Y. Hwang, "An improved search algorithm using incremental knowledge for continuous speech recognition,"InProc. IEEE Ins. Conf. Acoustics, Speech, and Signal Processing, vol. II, Minneapolis, MN, Apr, 1993, pp.307-310.
2. _, "improvements on the pronunciation prefix free search organization,'' in Proc. IEEE Ira. Conf. Acoustics, Speech, and Signal Processing, Atlanta, GA, May 1996, pp.133-136.
3. G. Antoniol, F. Brugnara, M. Cettolo, and M. Federico, "Language model representations for beam-search decoding,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing, vol. 1, Detroit, MI, Ma 1995, pp.5X8-591.
4. X. Aubert, "One pass cross word decoding lor large vocabularies based on a lexical tree search organization," inProc. Eur. Conf. Speech Communication and Technology, Budapest, Hungary, Sept. 1999, pp.1559-1562.
5. X. Aubert and H. Ney, "Large vocabulary continuous speech recognition using word graphs," inProc. IEEE In!. Conf. Acoustics, Speech, and Signal Processing, Detroit, MI, May 1995, pp.49-52.
6. X. Aubert, C. Dugast, H. Ney. and V. Steinbiss, "Large vocabulary, continuous speech recognition of wall street journal corpus," inProc. IEEE In!. Conf. Acoustics, Speech, and Signal Processing, vol. 2, Adelaide, Australia, Apr. 1994, pp.129-132.
7. S. Austin, R. Schwartz, and P. Placeway. "The forward-backward search algorithm," inProc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing, Toronto, Canada, May 1991, pp.697-700.
8. L. R. Bahl, F. Jelinek, and R. L. Mercer, "A maximum likelihood approach to continuous speech recognition," IEEE Trans. Pattern Anal. Machine Intel!., vol. PA M1-5. pp.179-190, Mar. 1983.
9. L. R. Bahl and F. Jelinek, "Decoding for channels with insertions, deletions and Substitutions with applications to speech récognition,"IEEE Trans. Inform. Theory, vol. IT-21, pp.404-411. July 1975.
10. J. K. Baker. "Stochastic modeling for automatic speech understanding,"InSpeech Recognition, D. R. Reddy, Ed. New York: Academic, 1975, pp.512-542.
11. R. E. Bellman, Dynamic Programming, Princeton, NJ: Princeton Univ. Press, 1957.
12. K. Beulen, S. Ortmanns, and C. Elting, "Dynamic programming search techniques for across-word modeling in speech recognition,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing, vol. 2, Phoenix, AZ, Mar. 1999, pp.609-612.
13. P. Beyeriein and M. L'llrich, "Hamming distance approximation for a fast log-likelihood computation for mixture densities,"InProc. Eur. Conf. Speech Communication and Technology, Madrid, Spain, Sept. 1995, pp.1083-1086.
14. E. Bocchieri, "Vector quantization for the efficient computation of continuous density likelihoods,"InProc. IEEE int Conf. Acoustics, Speech, and Signal Processing, Minneapolis, MN, Apr. 1993, pp.692-695.
15. J. S. Bridle, M. D. Brown, and R. M. Chamberlain, "An algorithm for connected word recognition,"InProc. IEEE int. Conf. Acoustics, Soeech, and Signal Processing, Paris, France, May 1982, pp.899-902.
16. R. Cardin, Y. Notmandin, and R. DeMori, "High performance connected digit recognition using codebook exponents,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing, vol. 1, San Francisco, CA, Mar. 1992, pp.505-08.
17. S. Delia Pietra, v. Delia Pietra, J. Gillett, J. Lafferty, H. Printz, and L. Ures, "Inference and estimation of a long-range trigram model,"InProc. 2nd Int. Coll. Grammatical Inference ana Applications, Alicante, Spain, Sept. 1994, pp.78-92.
18. N. Desmukh, A. Ganapathiraju, and J. Picone, "Hierarchical search for large vocabulary conversational speech recognition,"IEEE Signal Processing Mag., pp.84-107, Sept. 1999.
19. V. Digalakis and H. Murveit, "Genones: Optimizing the degree of mixture tying in a large vocabulary hidden Markov model-based speech recognizer,"InProc. IEEE Inf. Conf. Acoustics, Speech. and Signal Processing, vol. I, Adelaide, Australia, Apr. 1994. pp.537-540.
20. C. Dugast, R. Kneser, X. Aubert, S. Ortmanns, K. Beulen, and H. Ney, "Continuous speech recognition test and results for the NAB'94 corpus,"InProc. ARPA Spoken Language Technology Workshop, Austin, TX, Jan. 1995, pp.156-161.
21. L. D. Erman and V. R, Lesser, "The hearsay-!! speech understanding system,"InTrends in Speech Recognition, W. A. Lea, Ed. Englewood Cliffs, NJ: Prentice-Hall, 1980, pp.361-381.
22. L. Fissore, E. Giachin, P. Laface, and P. Massafra, "Using grammars in forward and backward search,"InProc. Eur. CotJ. Speech Communication and Technology, Berlin, Germany, Sept. 1993, pp.1525-1528.
23. G. D. Forney, "The viterbi algorithm," Proc. IEEE, vol. 61, pp.268-278, Mar. 1973.
24. J. L. Gauvain, L. F. Lamel, G. Adda, and M. Adda-Decker, "The L1MSI speech dictation system: Evaluation on the ARPA Wall Street Journal task,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing, vol. 1, Adelaide, Australia, Apr, 1994, pp.557-560.
25. R. Haeb-Umbach and H. Ney, "Improvements in time-synchronous beam search for lOOÜÜ-word continuous speech recognition,"IEEE Trans. Speech Audio Processing, voi. 2, pp.353-356, Apr. 1994.
26. M.-Y. Hwang, R. Rosenfeld, E. Thayer, R. Mosur, L. Chase, R. Weide, X. Huang, and F. Alleva, "Improving speech recognition performance via phone-dependent VQ codebooks and adaptive language models in SPHINX D,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing, vol. 1, Adelaide, Australia, Apr. 1994. pp.549-552.
27. F. Itakura, "Minimum prediction residual principle applied to speech recognition,"IEEE Trans. Acoust., Speech, Signal Processing, vol. ASSP-23, pp.67-72, Feb. 1975.
28. F. Jelinek, "Continuous speech recognition by statistical methods," Proc. IEEE, vol. 64, pp.532-556, Apr, 1976.'
29. D- H. Klatt, "Overview of the ARPA speech understanding project," hiTrends in Speech Recognition, W. A. Lea, Ed- Englewood Cliffs, NJ: Prentice-Hall, 1980, pp.249-271.
30. _, "SCRIBER and LAFS: Two new approaches to speech analysis,"InTrends in Speech Récognition, W. A. Lea, Ed. Englewood Cliffs, NJ: Prentice-Hall, 1980. pp.529-555.
31. B. M. Klovstad and L. F. Mondshein, "The CASPERS linguistic analysis system,"IEEE Trans. Acoust., Speech, Signal Processing, vol. ASSP-23, pp.118-123, Feb. 1975.
32. F. Kubala,'"Design of the 1994 CSR benchmark tests,"InProc. ARPA Spoken Language Technology Workshop, Austin, TX, Jan. 1995. pp.41-46.
33. F. Kiibala, A. Anastasakos, j. Makhoul, L. Xguyen, and R. Schwanz, "Comparative experiments on large vocabulary speech récognition,"InProc. IEEE Int. Cor.f. Acoustics, Speech, and Signal Processing, vol. 1, Adelaide, Australia, Apr. 1994, pp.561-564.
34. R. Kulm and R. De Mori, "A cache-based natural language model for speech recognition,"IEEE Trans. Pattern Anal. Machine Intel!., vol. 12, pp.570-583, June 1990.
35. T. Kühn, P. Fetter, A. Kaltenmeier, and P. Regel-Brietzmann, "DP-based wordgraph pruning,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing, Atlanta, GA, May 1996, pp.461-464.
36. R. Lacouture and Y. Normandin, "Efficient lexical access strategies,"InProc. Eia: Conf. Speech Communication and Technology, Berlin, Germany, Sept. 1993, pp.1537-1540.
37. C. H. Lee and L. R. RaDiner, "A frame-synchronous network search algorithm for continuous word recognition,"IEEE Trans. Acoust., Speech, Signal Processing, vol. 37, pp.1649-1658, Nov. 1939.
38. S. E. Levinson, L. R. Rabiner, and M. M. Sondhi, "An introduction to the application of the theory of probabilistic functions of a Markov process to automatic speech recognition," Beli Syst. Tech. J., vol. 62, pp.1035-1074, Apr. 1983.
39. A. Ljolje, M. Riiey, D. Hindle, and F. Pereira, "The AT&T 60000 word peech-to-text system,"InProc. ARPA Spoken Language Systems Technology Workshop, Austin, TX, Jan. 1995, pp.162-165.
40. J B. T. Lowerre and R. Reddy, "The HARPY speech understanding system,"InTrends in Speech Recognition, W. A. Lea, Ed. Englewood Cliffs, NJ: Prentice-Hall, 1980, pp.340-360.
41. K. Mehlhorn, Data Structures and Algorithms I: Sorting and Searching. Berlin, Germany: Springer-Verlag, 1984.
42. H. Murveit, J. Butzberger, v. Digalakis, and VI. Weintraub, "Large-vocabulary dictation using SRI's decipher speech recognition system: Progressive-search techniques,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing, vol. 2, Minneapolis, MN, Apr. 1993, pp.319-322.
43. C. S. Myers and L. R. Rabiner, "Connected digit recognition using a level-building DTW algorithm,"IEEE Trans. Acoust., Speech, Signal Processing, vol. ASSP-29, pp.351-363, June 1981.
44. H. Ney, "Connected utterance recognition using dynamic programming,"InFortschritte der Akustik-FASE!DAGA'82, Federation of Acoustic Societies of Europe!Deutsche Arbeitsgemeinschaft für Akustik, Göttingen, Germany, Sept. 1982, pp.915-918.
45. _, "The use of a one-stage dynamic programming algorithm for connected word recognition,"IEEE Trans. Acoust., Speech, Signal Processing, vol. ASSP-32, pp.263-271, Apr. 1984.
46. _, "Acoustic modeling of phoneme units for continuous speech recognition,"InProc. 5th Eur. Signal Processing Conf., Barcelona, Spain, Sept. 1990, pp.65-72.
47. _, "Dynamic programming parsing for context free grammars in continuous speech recognition,"IEEE Trans. Signal Processing, vol. 39, pp.336-341, Feb. 1991.
48. H. Ney, "Search strategies for large-vocabulary continuous-speech recognition,"InSpeech Recognition ami Coding-New Advances and Trends, A. j. Rufaio Ayuso and J. M. Lopez Soier, Eds. Berlin, Germany: Springer, 1995, pp.210-225.
49. H. Ney.'R, Haeb-Umbach, B.-H. Tran, and M. Oerder. "Improvements in beam search for 10 000-word continuous speech recognition,"InProc. IEEE Int. Conf. Acoustics. Speech, and Signal Processing, San Francisco, CA, Mar. 1992, pp.13-16.
50. H. Ney, D. Mergel, A. Noli, and A. Paeseler, "Data driven organization of the dynamic programming beam search for continuous speech recognition,1' IEEE Trans. Signal Processing, vol. 40, pp.272-281, Feb. 1992.
51. H. Ney and S. Ortmanns, "Extensions to the word graph method for large vocabulary continuous speech recognition,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing, vol. 3, Munich, Germany, Apr. 1997, pp.1787-1790.
52. L. Nguyen and R. Schwanz, "Single-tree method for grammar-di- ' reeled search,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signa! Processing, vol. 2, Phoenix, AZ, Mar. 1999, pp.613-616.
53. J J. J. Odell, v. Valtchev, P. C. Woodland, andS J. Young, "A one-pass decoder design for large vocabulary recognition,"InProc. ARPA Spoken Language Technology Workshop, Plainsboro, NJ, Mar. 1994, pp.405-410.
54. J. J. Odell, "The use of context in large vocabulary speech recognition," Ph.D. dissertation, Elect, Eng. Dept, Cambridge Univ., Cambridge, U.K. Mar. 1995.
55. M. Oerder and H. Ney, "Word graphs: An efficient interface between continuous speech recognition and language understanding,"InProc. IEEE Int. Conf. Acoustics, Speecit, and Signal Processing, vol. 2, Minneapolis, MN, Apr. 1993, pp.119-122.
56. S. Ortmarjis, A. Eiden, and H. Ney, "Improved lexical tree search for large vocabulary speech recognition,"InProc. IEEE In!. Conf. Acoustics, Speech, and Signal Processing, Seattle, WA, May 1998, pp.817-820.
57. S. Ortmanns, A. Eiden, H. Ney, and N. Coenen, "Look-ahead techniques for fast beam search,"InProc. Int. Conf. Acoustics, Speech and Signal Processing, vol. 3, Munich, Germany, Apr. 1997, pp.1783-1786.
58. S. Ortmanns and H. Ney, "Experimental analysis of the search space for 20 000-word speech recognition,"InProc. 4th Ear. Conf. Speech Communication and Technology, Madrid, Spain, Sept. 1995, pp.901-904.
59. _, "Time-conditioned search for large vocabulary continuous speech recognition,"IEEE Trans. Speech Audio Processing, to be published.
60. S. Ortmanns, H. Ney, and X. Auhert, "A word graph algorithm for large vocabulary continuous speech recognition," Coniput., Speech Lam,., vol. 11, pp.43-72, Jan. 1997.
61. S. Ortmanns, H. Ney, and A. Eiden, ''Language-model look-ahead for large vocabulary speech recognition,"InProc. Int. Conf. Spoken Language Processing, Philadelphia, PA, Oct. 1996, pp.2095-2098.
62. S. Ortmanns, H. Key, and T. Firzlaff, "Fast likelihood computation methods for continuous mixture densities in large vocabulary speech recognition,"InProc. 5th Eur. Conf. Speech Communication and Technology, Rhodes, Greece, Sept. 1997, pp.143-146.
63. S. Ortmanns, H. Ney, F. Seide, and I. Lindam, "A comparison of time conditioned and word conditioned search techniques for large vocabulary speech recognition,"InInt. Conf. Spoken Language Processing, Philadelphia, PA, Oct. 1996, pp.2091-2094.
64. D. B. Paul, "Algorithms for an optimal A"' search and linearizing the search in the stack decoder,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signa! Processing, Toronto, Canada, May 1991, pp.693-996.
65. _, "An efficient A stack decoder algorithm for continuous speech recognition with a stochastic language model,"InProc. IEEE In!. Conf. Acoustics, Speech, and Signal Processing, San Francisco, CA, Mar. 1992, pp.25-28.
66. L. R. Rabiner, "A tutorial on hidden Markov models and selected applications in speech recognition," Proc. IEEE, vol. 77, pp.257-2X6, Feb. 1989.
67. S. Renais and M. Hochberg, "Efficient search asing posterior phone probability estimates,"InProc. IEEE Int. Conf. Acoustics, Speech, Signal Processing, Detroit, MI, May 1995, pp.596-599.
68. S. Renais and M. Hochberg, "Efficient evaluation of the LVCSR search space using the NOWAY decoder,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing, Atlanta, GA, May 1996, pp.149-152.
69. H. Sakoe. "Two-level DP matching-A dynamic programming-based pattern matching algorithm for connected word recognition,"IEEE Trans. Acoust., Speech, Signal Processing, vol. ASSP-27, pp.588-595, Dec. 1979.
70. H. Sakoe and S. Chiba, "A dynamic programming approach to continuous speech recognition,"InProc. 7th Im. Congr. Acoustics, Budapest, Hungary, Aug. 1971, Paper 20 C 13, pp.65-68.
71. R- Schwanz and S. Austin, "A comparison of several approximate algorithms for finding multiple (.Y-best) sentence hypotheses,"InProc. IEEE Ira. Conf. Acoustics, Speech, and Signal Processing. Toronto, Canada, May 1991, pp.701-704.
72. A. Sixtus and S. Ortmanns, "High quality word graphs using forward-backward pruning,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing, vol. 2, Phoenix, AZ, Mar. 1999, pp.593-596.
73. F. K. Soong and E.-F. Huang, "A tree-trellis fast search for finding the A'-best sentence hypotheses,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing, Toronto, Canada, May 1991, pp.705-708.
74. V. Steinbiss, "A search organization for large vocabulary recognition based upon JV-best decoding,"InProc. 2nd Eur. Conf. Speech Communication and Technology, vol. 3, Geneva, Itaiy, Sept. 1991, pp.1217-1220.
75. _, personal communication. Philips Research Laboratories, Aachen, Germany, June 1992.
76. V. Steir.biss, H. Key, R. Haeb-Umbads, B.-H. Tran, U. Essen, and R. Kneser e: u!., "The Philips researcn system lor large-vocabulary continuous-speech recognition,"I P roc. 3rd Kur. Con}. Speech Communication und Technolu, Berlin, Germany, Sept, 1991, pp.2125-2128
77. V. Steinbiss, B.-H. Trau, and H. Ney, "Improvements in beam searcn."In Int. Conf. Spoken Laicguaze Processing. Yokohama, Japan, Sept. 1994, pp.1353-1358.
78. C. Tillmann and H. Ney, "Word triggers and the EM algorithm,"InACL Spécial Interest Group Workshop Cumpuiational Saturai Limpliage Learning, Madrid, Spain. July 1997, pp.117-124.
79. V. \I. Velichko and N. G. Zagoruyko, "Automatic recognition of 200 words,"Int..J. Man-Machine Stiidie;;. vol. 2, pp.223-234, June 1970.
80. T. K. Vintsyuk, "Speech discrimination by dynamic programming," Kibernelika, vol. 4, pp.81-88, Jan.-Feb. 1968.
81. _, "Elementwisc recognition of continuous speech composed of words from a specified dictionary,'" Xiberneiika, vol. 7, pp.133-143, Mar./Apr. 1971.
82. J. J. Wolf and W. A. Woods, "The HWIM speech understanding system,"InTrends in Speech Recognition, W. A. Lea, Ed. Englewood Cliffs, NJ: Prentice-Hal!,' 1980, pp.316-393.
83. [S3]D. Wood, Darn Structure-;. Algorithms and Performance. Reading, MA: Addison-Wesley, 1993,
84. P. C. Woodland, J. J. Odd], v. Valiech, and S. J. Young, "Large vocabulary continuous speech lecognition using HTK,"InP roc. IEEE In!. Conf. Aco.istu'S, Speech, and 5'/,e/(j/ Processing, vol. 2, Adelaide, Australia, Apr. 1994, pp.,25-128.
85. S. J. Young, J. J. Odell, and P. C. Woodland, "Tree-based state tying for high accuracy acoustic modeling,"InARPA Human Language Technology Workshop. San Vlateo. CA: Morgan Kaufmann, Mar. 1994, pp.286-291.
86. Q. Zhou and W. Chou, "An appioach to continuous speech recognition baseci on a layered self-adjusting decoding graph,"InProc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing, Munich. Germany, Apr. 1997, pp.1779 1782.
87. H. Ney, "Search strategies for large-vocabulary continuous-speech recognition," NATO Advanced Studies Institute, Bubicn. Spain, June-July 1993.