Discriminative Training for Segmental Minimum Bayes-Risk Decoding

Download: PDF, slides.

“Discriminative Training for Segmental Minimum Bayes-Risk Decoding” by V. Doumpiotis, S. Tsakalidis, and W. Byrne. In IEEE Conference on Acoustics, Speech and Signal Processing, 2003, IEEE.


A modeling approach is presented that incorporates discriminative training procedures within segmental Minimum Bayes-Risk decoding (SMBR). SMBR is used to segment lattices produced by a general automatic speech recognition (ASR) system into sequences of separate decis ion problems involving small sets of confusable words. Acoustic models specialized to discriminate between the competing words in these classes are then applied in subsequent SMBR rescoring passes. Refinement of the search space that allows the use of specialized discriminative models is shown to be an improvement over rescoring with conventionally trained discriminative models.

Download: PDF, slides.

BibTeX entry:

   author = {V. Doumpiotis and S. Tsakalidis and W. Byrne},
   title = {Discriminative Training for Segmental Minimum {B}ayes-Risk
   booktitle = {IEEE Conference on Acoustics, Speech and Signal Processing},
   pages = {(4 pages)},
   year = {2003},
   organization = {IEEE}

Back to Bill Byrne publications.