Comparing linear and non-linear transformation of speech

Authors:
Larbi Mesbahi;Vincent Barreaud;Olivier Boeffard
Affiliations:
IRISA, ENSSAT, University of Rennes 1, Lannion, France;IRISA, ENSSAT, University of Rennes 1, Lannion, France;IRISA, ENSSAT, University of Rennes 1, Lannion, France
Venue:
SSIP '09/MIV'09 Proceedings of the 9th WSEAS international conference on signal, speech and image processing, and 9th WSEAS international conference on Multimedia, internet & video technologies
Year:
2009

Citing 2
Cited 0

Speech spectrum conversion based on speaker interpolation and multi-functional representation with weighting by radial basis function networks

Speech Communication - Special issue: voice conversion: state of the art and perspectives
Reformulated radial basis neural networks trained by gradient descent

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper aims to study voice conversion using linear and non-linear transform systems, based on respectively the Gaussian Mixture Models, GMM, and the Radial Basis function, RBF.We compare on an identical speech database both proposed approaches. We insist in particular on the objective measures of the transformation, in the case that we have not enough data recorded for the target speaker. We show for databases containing only one and two speech sentences, that the non-linear transform (RBF) gives weaker distortion scores than the GMM.