Restructuring Gaussian mixture density functions in speaker-independent acoustic models

Speech Communication

Volumn 36, Issue 3-4, 2002, Pages 277-289

  Nakamura, A ^a  

^a NTT Communication Science Laboratories   (Japan)

Author keywords

Acoustic model; Frame level error; Gaussian mixture density function; Modeling mismatch; Speech recognition

Indexed keywords

GAUSSIAN NOISE (ELECTRONIC); PROBABILITY DENSITY FUNCTION; SPEECH PROCESSING;

GAUSSIAN MIXTURE DENSITY FUNCTION;

CONTINUOUS SPEECH RECOGNITION;

EID: 0036497676     PISSN: 01676393     EISSN: None     Source Type: Journal    
DOI: 10.1016/S0167-6393(00)00087-X     Document Type: Article

References (17)

1. Maximum mutual information estimation of hidden Markov models
2. Segmental GPD training of HMM based speech recognizer
3. Unified techniques for vector quantisation and hidden Markov modeling using semi-continuous models
4. On the semi-continuous hidden Markov modeling
5. Discriminative learning for minimum error classification
6. Pattern recognition using a family of design algorithms based upon the generalized probabilistic descent method
7. Variable order N-gram generation by word class splitting and consecutive word grouping
8. A speech and language database for speech translation research
9. Predicting speech recognition performance
10. Japanese speech databases for robust speech recognition
11. HMM topology design using maximum likelihood successive state splitting
12. Using multiple vector quantization and semicontinuous hidden Markov models for speech recognition
13. Multiple VQ hidden Markov modeling for speech recognition
14. Delayed decision beam search for continuous speech recognition
15. Spontaneous dialogue speech recognition using cross-word context constrained word graphs
16. Decoding optimal state sequence with smooth state likelihoods