This paper describes Brno University of Technology (BUT) sys- tem for the Interspeech 2010 Paralinguistic Challenge. Our submitted systems for the Age- and Gender-Sub-Challenges employ fusions of several sub-systems. We make use of our own acoustic frame-based feature sets, as well as the provided utterance-based acoustic, prosodic and voice quality features. Modeling is based on Gaussian Mixture Models (GMM) and Support Vector Machines (SVM), followed by linear Gaussian backends and logistic regression-based fusion. For a single sub- system, we obtain improvement of about 2% absolute, for both tasks, on the development-set. Our ?nal fusion results in nearly 9% absolute improvement for the Age task and about 4.5% for the Gender task on the development set. On the ?nal test set we obtain 3.5% and 2% absolute improvement, respectively.
Cite as: Kockmann, M., Burget, L., Černocký, J. (2010) Brno university of technology system for interspeech 2010 paralinguistic challenge. Proc. Interspeech 2010, 2822-2825, doi: 10.21437/Interspeech.2010-746
@inproceedings{kockmann10b_interspeech,
author={Marcel Kockmann and Lukáš Burget and Jan Černocký},
title={{Brno university of technology system for interspeech 2010 paralinguistic challenge}},
year=2010,
booktitle={Proc. Interspeech 2010},
pages={2822--2825},
doi={10.21437/Interspeech.2010-746}
}