We introduce a set of speaker dependent features derived from the positions of vowels in Mel-Frequency Cepstral Coefficient (MFCC) space relative to a reference vowel. The MFCCs for a particular speaker are transformed using simple operations into features that can be used to classify vowels from a common reference point. Classification performance of vowels using Gaussian Mixture Models (GMMs) is significantly improved, regardless of which vowel is used as the target among /A/, /i/, /u/, or /ยด/. We discuss how this technique can be applied to assess pronunciation with respect to vowel structure rather than agreement with absolute position in MFCC space.
Cite as: Peabody, M., Seneff, S. (2010) A simple feature normalization scheme for non-native vowel assessment. Proc. Second Language Studies: Acquisition, Learning, Education and Technology (L2WS 2010), paper O2-2
@inproceedings{peabody10_l2ws,
author={Mitchell Peabody and Stephanie Seneff},
title={{A simple feature normalization scheme for non-native vowel assessment}},
year=2010,
booktitle={Proc. Second Language Studies: Acquisition, Learning, Education and Technology (L2WS 2010)},
pages={paper O2-2}
}