Improved modeling and efficiency for automatic transcription of Broadcast News

Speech Communication

Volumn 37, Issue 1-2, 2002, Pages 133-158

  Sankar, Ananth ^a,b   Ramana Rao Gadde, Venkata ^a   Stolcke, Andreas ^a   Weng, Fuliang ^a  

^a SRI INTERNATIONAL   (United States)

^b NUANCE COMMUNICATIONS   (United States)

Author keywords

Acoustic adaptation; Acoustic modeling; Language modeling; Lattice based decoding

Indexed keywords

ALGORITHMS; AUDIO ACOUSTICS; DATA REDUCTION; MATHEMATICAL MODELS; SPEECH RECOGNITION;

AUTOMATIC TRANSCRIPTION;

RADIO BROADCASTING;

EID: 0036567791     PISSN: 01676393     EISSN: None     Source Type: Journal    
DOI: 10.1016/S0167-6393(01)00063-2     Document Type: Article

References (50)

1. The forward-backward search algorithm
2. A latent semantic analysis framework for large-span language modeling
3. Language model adaptation using mixtures and an exponentially decaying cache
4. Maximum likelihood from incomplete data via the EM algorithm
5. High-accuracy large-vocabulary speech recognition using mixture tying and consistency modeling
6. Speaker adaptation using constrained reestimation of Gaussian mixtures
7. Genones: Generalized mixture tying in continuous hidden Markov model-based speech recognizers
8. Speaker adaptation using combined transformation and Bayesian methods
9. CSR corpus development
10. Maximum a posteriori estimation for multivariate Gaussian mixture observations of Markov chains
11. SWITCHBOARD: Telephone speech corpus for research and development
12. The population frequencies of species and the estimation of population parameters
13. Quantizing mixture weights in a tied-mixture HMM
14. Acoustic clustering and adaptation for robust speech recognition
15. Predicting unseen triphones with Senones
16. Modeling long distance dependencies in language: Topic mixtures vs. dynamic cache models
17. Maximum-likelihood estimation for mixture multivariate stochastic observations of Markov chains
18. Statistical language modeling using a variable context length
19. The 1996 BBN Byblos Hub-4 transcription system
20. Flexible speaker adaptation using maximum likelihood linear regression
21. An algorithm for vector quantizer design
22. Weighted determinization and minimization for large vocabulary speech recognition
23. Large-vocabulary dictation using SRI's DECIPHER (TM) speech recognition system: Progressive-search techniques
24. A comparative study of speaker adaptation techniques
25. A word graph algorithm for large vocabulary, continuous speech recognition
26. Experiments with a Gaussian merging-splitting algorithm for HMM training for speech recognition
27. A new look at HMM parameter tying for large vocabulary speech recognition
28. Robust HMM estimation with Gaussian merging-splitting and tied-transform HMMs
29. Parameter tying and Gaussian clustering for faster, better, and smaller speech recognition
30. Stochastic matching for robust speech recognition
31. A maximum-likelihood approach to stochastic matching for robust speech recognition
32. Acoustic modeling for the SRI Hub4 partitioned evaluation continuous speech recognition system
33. Development of SRI's 1997 Broadcast News transcription system
34. SRI's 1998 Broadcast News system - Toward faster, smaller, better speech recognition
35. A comparison of several approximate algorithms for finding multiple (N-best) sentence hypotheses
36. Scalable backoff language models
37. Using story topics for language model adaptation
38. Specification of the 1996 Hub4 broadcast news evaluation
39. Hub-4 language modeling using domain interpolation and data clustering
40. Efficient lattice representation and generation
41. New developments in lattice-based search strategies in SRI's H4 system
42. Large vocabulary continuous speech recognition using HTK
43. The 1997 HTK Broadcast News transcription system
44. The use of state tying in continuous speech recognition