Shared latent subspace modelling within Gaussian-Binary Restricted Boltzmann Machines for NIST i-Vector Challenge 2014
Machine Learning
2015-03-19 v1 Neural and Evolutionary Computing
Sound
Machine Learning
Abstract
This paper presents a novel approach to speaker subspace modelling based on Gaussian-Binary Restricted Boltzmann Machines (GRBM). The proposed model is based on the idea of shared factors as in the Probabilistic Linear Discriminant Analysis (PLDA). GRBM hidden layer is divided into speaker and channel factors, herein the speaker factor is shared over all vectors of the speaker. Then Maximum Likelihood Parameter Estimation (MLE) for proposed model is introduced. Various new scoring techniques for speaker verification using GRBM are proposed. The results for NIST i-vector Challenge 2014 dataset are presented.
Keywords
Cite
@article{arxiv.1503.05471,
title = {Shared latent subspace modelling within Gaussian-Binary Restricted Boltzmann Machines for NIST i-Vector Challenge 2014},
author = {Danila Doroshin and Alexander Yamshinin and Nikolay Lubimov and Marina Nastasenko and Mikhail Kotov and Maxim Tkachenko},
journal= {arXiv preprint arXiv:1503.05471},
year = {2015}
}
Comments
5 pages, 3 figures, submitted to Interspeech 2015