ISCA Archive Interspeech 2005
ISCA Archive Interspeech 2005

A hybrid ANN/DBN approach to articulatory feature recognition

Joe Frankel, Simon King

Artificial neural networks (ANN) have proven to be well suited to the task of articulatory feature (AF) recognition. However, one drawback with an ANN approach is that features are assumed to be statistically independent. We address this by using ANNs to provide virtual evidence to a dynamic Bayesian network (DBN). This gives a hybrid ANN/DBN model and allows modelling of inter-feature dependencies. We demonstrate significant increases in AF recognition accuracy from modelling dependencies between features, and present the results of embedded training experiments in which a set of asynchronous feature changes are learned. Furthermore, we report on the application of a Viterbi training scheme in which we alternate between realigning the AF training labels and retraining the ANNs.

doi: 10.21437/Interspeech.2005-150

Cite as: Frankel, J., King, S. (2005) A hybrid ANN/DBN approach to articulatory feature recognition. Proc. Interspeech 2005, 3045-3048, doi: 10.21437/Interspeech.2005-150

  author={Joe Frankel and Simon King},
  title={{A hybrid ANN/DBN approach to articulatory feature recognition}},
  booktitle={Proc. Interspeech 2005},