An alternative approach of finding competing hypotheses for better minimum classification error training

ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

Volumn 1, Issue , 2002, Pages

  Tam, Yik Cheung ^a   Mak, Brian ^a  

^a HONG KONG UNIVERSITY OF SCIENCE AND TECHNOLOGY   (Hong Kong)

Author keywords

[No Author keywords available]

Indexed keywords

BEAM SEARH ALGORITHM; COMPETING HYPOTHESIS; GENERALIZED PROBABILISTIC DESCENT ALGORITHM; HIDDEN MARKOV MODELS; MINIMUM CLASSIFICATION ERROR TRAINING; N-BEST ALGORITHM; SIGMOID FUNCTION; WORD ERROR RATES;

BIT ERROR RATE; DATABASE SYSTEMS; ERROR ANALYSIS; LEARNING ALGORITHMS; LEARNING SYSTEMS; MARKOV PROCESSES; MAXIMUM LIKELIHOOD ESTIMATION; OPTIMIZATION; PARAMETER ESTIMATION; PROBABILITY DENSITY FUNCTION;

SPEECH RECOGNITION;

EID: 0742302023     PISSN: 15206149     EISSN: None     Source Type: Conference Proceeding    
DOI: None     Document Type: Conference Paper

References (8)

1. Minimum classification error rate methods for speech recognition
2. Minimum error rate training based on N-best string models
3. An application of discriminative feature extraction to filter-bank-based speech recognition
4. Development of an asynchronous multi-band system for continuous speech recognition
5. Tree-trellis based fast search for the finding of the N-best sentence hypotheses in continuous speech recognition
6. He N-best algorithm: An efficient and exact procedure for finding the N most likely sentence hypotheses
7. The AURORA experimental framework for the performance evaluations of speech recognition systems under noisy conditions
8. A database for speaker-independent digit recognition