Xue Wang, Louis F. M. ten Bosch & Louis C. W. Pols
(Postscript (225k) and RTF (167k)
versions are available)
ABSTRACT
This paper presents research on integrating context-dependent durational
knowledge into HMM-based speech recognition. The first part of the paper
presents work on obtaining relations between the parameters of the context-free
HMMs and their durational behaviour, in preparation for the context-dependent
durational modelling presented in the second part. Duration integration is
realised via rescoring in the post-processing step of our N-best monophone
recogniser. We use the multi-speaker TIMIT database for our analyses.
- introduction
- dpdf of standard hmm
- Obtaining the dpdf of the whole HMM
- Analysis of whole-model dpdf
- ml-training constrained with durational statistics
- analysis of context-dependent durational statistics
- integration of CD-duration models in post-processing
- Word-juncture modelling
- Duration score
- Re-scoring
- discussion
- REFERENCES