Title :
Speaker verification using various prosodic kernels
Author :
Szymon Drgas;Dariusz Zamorski;Adam Dąbrowski
Author_Institution :
Poznan University of Technology, Chair of Control and Systems Engineering, ul. Piotrowo 3a, PL-60 965 Poznan Poland
Abstract :
In this article text independent speaker verification problem is considered. The approach in which SVM and prosodic features are employed has been chosen. Prosodic features are modelled by Legendre polynomials. In order to map a sequence of prosodic vectors to fixed dimensional vector three kernels were used: VQ kernel, GMM supervector kernel and Fisher kernel. These three kernels were compared and its combination was evaluated.
Keywords :
"Kernel","Training","Vectors","Support vector machines","Speech","Polynomials","Feature extraction"
Conference_Titel :
Signal Processing Algorithms, Architectures, Arrangements, and Applications Conference Proceedings (SPA), 2011
Print_ISBN :
978-1-4577-1486-3
