13th Annual Conference of the International Speech Communication Association

Portland, OR, USA
September 9-13, 2012

A Comparison of Classification Paradigms for Speaker Likeability Determination

Nicholas Cummins, Julien Epps, Jia Min Karen Kua

School of Electrical Engineering and Telecommunications, The University of New South Wales, Sydney, NSW, Australia
ATP Research Laboratory, National ICT Australia (NICTA), Eveleigh, Australia

In this paper we investigate the performance of different classification paradigms, testing each with a range of acoustic features, to find a system that is well suited to speaker likeability classification. We introduce a Sparse Representation Classifier for paralinguistic classification and explore the role of training data selection for a GMM classifier. Results demonstrate that (1) Single dimensional features of pitch direction, shimmer and spectral roll-off were the most suitable features found when testing on the development set but we were unable to reproduce their performance in the final classification task, (2) Using UBM training data selection increased accuracy of MFCC's and (3) Sparse Representation showed promise as a paralinguistic classifier with results comparable to that of SVM.

Index Terms: Likeability, single and multi-dimensional feature selection, Sparse Representation Classifier, UBM data selection.

Full Paper

Bibliographic reference.  Cummins, Nicholas / Epps, Julien / Kua, Jia Min Karen (2012): "A comparison of classification paradigms for speaker likeability determination", In INTERSPEECH-2012, 282-285.