Paper: Speaker Recognition With Mixtures Of Gaussians With Sparse Regression Matrices

ACL ID N04-2010
Title Speaker Recognition With Mixtures Of Gaussians With Sparse Regression Matrices
Venue Human Language Technologies
Session Student Session
Year 2004
Authors

When estimating a mixture of Gaussians there are usually two choices for the covariance type of each Gaussian component. Either diag- onal or full covariance. Imposing a struc- ture though may be restrictive and lead to de- graded performance and/or increased compu- tations. In this work, several criteria to esti- mate the structure of regression matrices of a mixture of Gaussians are introduced and eval- uated. Most of the criteria attempt to estimate a discriminative structure, which is suited for classification tasks. Results are reported on the 1996 NIST speaker recognition task and performance is compared with structural EM, a well-known, non-discriminative, structure- finding algorithm.