A combined strategy of noise-adaptive training (NAT) and discriminative-based adaptation is proposed for effective migration of speech recognition systems to other noisy environments. NAT is an effective approach for real-field applications, but does not satisfy the minimum classification error (MCE) criterion for the recognition process and adapts poorly to new environments. The proposed method makes up for the weak points in discriminative adaptation strategies, and presents a new method for improving the MCE approach. Using this new method, experimental results show that the speech recognition system can successfully be migrated to other environments using specific-condition data of the target environment.
Cite as: Kang, B.-O., Jung, H.-Y., Lee, Y.-K. (2007) Discriminative noise adaptive training approach for an environment migration. Proc. Interspeech 2007, 2085-2088, doi: 10.21437/Interspeech.2007-564
@inproceedings{kang07_interspeech,
author={Byung-Ok Kang and Ho-Young Jung and Yun-Keun Lee},
title={{Discriminative noise adaptive training approach for an environment migration}},
year=2007,
booktitle={Proc. Interspeech 2007},
pages={2085--2088},
doi={10.21437/Interspeech.2007-564}
}