ISCA Archive Eurospeech 1997
ISCA Archive Eurospeech 1997

A comparative study of methods for phonetic decision-tree state clustering

H. J. Nock, M. J. F. Gales, Steve J. Young

Phonetic decision trees have been widely used for obtaining robust context-dependent models in HMM-based systems. There are five key issues to consider when constructing phonetic decision trees: the alignment of data with the chosen phone classes; the quality of the modelling of the underlying data; the choice of partitioning method at each node; the goodness-of-split criterion and the method for determining appropriate tree sizes. A popular existing method usesefficient but crude approximatemethods for each of these. This paper introduces and evaluates more detailed alternatives to the standard approximations.


doi: 10.21437/Eurospeech.1997-55

Cite as: Nock, H.J., Gales, M.J.F., Young, S.J. (1997) A comparative study of methods for phonetic decision-tree state clustering. Proc. 5th European Conference on Speech Communication and Technology (Eurospeech 1997), 111-114, doi: 10.21437/Eurospeech.1997-55

@inproceedings{nock97_eurospeech,
  author={H. J. Nock and M. J. F. Gales and Steve J. Young},
  title={{A comparative study of methods for phonetic decision-tree state clustering}},
  year=1997,
  booktitle={Proc. 5th European Conference on Speech Communication and Technology (Eurospeech 1997)},
  pages={111--114},
  doi={10.21437/Eurospeech.1997-55},
  issn={1018-4074}
}