We propose a new algorithm called Generalized Discriminative Feature Transformation (GDFT) for acoustic models in speech recognition. GDFT is based on Lagrange relaxation on a transformed optimization problem. We show that the existing discriminative feature transformation methods like feature space MMI/MPE (fMMI/MPE), region dependent linear transformation (RDLT), and a non-discriminative feature transformation, constrained maximum likelihood linear regression (CMLLR) are special cases of GDFT. We evaluate the performance of GDFT for Iraqi large vocabulary continuous speech recognition.
Cite as: Hsiao, R., Schultz, T. (2009) Generalized discriminative feature transformation for speech recognition. Proc. Interspeech 2009, 664-667, doi: 10.21437/Interspeech.2009-232
@inproceedings{hsiao09_interspeech,
author={Roger Hsiao and Tanja Schultz},
title={{Generalized discriminative feature transformation for speech recognition}},
year=2009,
booktitle={Proc. Interspeech 2009},
pages={664--667},
doi={10.21437/Interspeech.2009-232}
}