Speaker recognition by means of a combination of linear and nonlinear predictive models

TL;DR

This paper enhances speaker recognition accuracy by combining linear and nonlinear predictive models with LPCC features, showing significant error rate reductions and proposing an efficient computational algorithm.

Contribution

It introduces a novel combination of predictive models with LPCC for improved speaker recognition performance.

Findings

Error rate reduced from 6.31% to 3.68% with nonlinear residuals.

Linear prediction residuals improve accuracy by 2.63%.

Proposes an efficient algorithm for computational reduction.

Abstract

This paper deals the combination of nonlinear predictive models with classical LPCC parameterization for speaker recognition. It is shown that the combination of both a measure defined over LPCC coefficients and a measure defined over predictive analysis residual signal gives rise to an improvement over the classical method that considers only the LPCC coefficients. If the residual signal is obtained from a linear prediction analysis, the improvement is 2.63% (error rate drops from 6.31% to 3.68%) and if it is computed through a nonlinear predictive neural nets based model, the improvement is 3.68%. An efficient algorithm for reducing the computational burden is also proposed.