This work is motivated by our earlier study which shows that on explicit pitch normalization the childrens speech recognition performance on the adults speech trained models improves as a result of reduction in the pitch-dependent distortions in the spectral envelope. In this paper, we study the role of spectral smoothing in context of childrens speech recognition. The spectral smoothing has been effected in the feature domain by two approaches viz., modification of bandwidth of the filters in the filterbank and cepstral truncation. In conjunction, both approaches give significant improvement in the childrens speech recognition performance with 57% relative improvement over the baseline. Also, when combined with the widely used vocal tract length normalization (VTLN), these spectral smoothing approaches result in an additional 25% relative improvement over the VTLN performance for childrens speech recognition on the adults speech trained models.
Cite as: Ghai, S., Sinha, R. (2009) Exploring the role of spectral smoothing in context of children's speech recognition. Proc. Interspeech 2009, 1607-1610, doi: 10.21437/Interspeech.2009-209
@inproceedings{ghai09_interspeech,
author={Shweta Ghai and Rohit Sinha},
title={{Exploring the role of spectral smoothing in context of children's speech recognition}},
year=2009,
booktitle={Proc. Interspeech 2009},
pages={1607--1610},
doi={10.21437/Interspeech.2009-209}
}