Command-response model is one of the quantitative models capable of illustrating the voice fundament frequency (F0) characteristics of different languages. In recent years, the model has been adapted to tone languages including Cantonese. This paper presents an automatic optimization procedure to enhance parameter extraction of command-response model for short phrases of Cantonese. We conduct a parameter extraction test on 128 speech segments produced by three speakers, all of which are modeled by single phrase command in the command-response model. The extracted parameters attained accuracy compatible with that obtained from the conventional manual estimation.
Cite as: Ng, R.W.M., Lee, T., Gu, W. (2006) Towards automatic parameter extraction of command-response model for Cantonese. Proc. Interspeech 2006, paper 1363-Thu1FoP.1, doi: 10.21437/Interspeech.2006-591
@inproceedings{ng06_interspeech,
author={Raymond W. M. Ng and Tan Lee and Wentao Gu},
title={{Towards automatic parameter extraction of command-response model for Cantonese}},
year=2006,
booktitle={Proc. Interspeech 2006},
pages={paper 1363-Thu1FoP.1},
doi={10.21437/Interspeech.2006-591}
}