Variance-Spectra based Normalization for I-vector Standard and Probabilistic Linear Discriminant Analysis

• http://www.fit.vutbr.cz/research/groups/speech/publi/2012/bousquet_odyssey2012_157-164-09.pdf

i-vectors, probabilistic linear discriminant analysis, speaker recognition

This paper is on various i-vector normalizations for speaker recognition using standard and probabilistic Linear Discriminant Analysis (LDA and PLDA)

I-vector extraction and Probabilistic Linear Discriminant Analysis (PLDA) has become the state-of-the-art configuration for speaker verification. Recently, Gaussian-PLDA has been improved by a preliminary length normalization of i-vectors. This normalization, known to increase the Gaussianity of the i-vector distribution, also improves performance of systems based on standard Linear Discriminant Analysis (LDA) and "two-covariance model" scoring. We propose in this paper to replace length normalization by two new techniques based on total, between- and within-speaker variance spectra 1. These "spectral" techniques both normalize the i-vectors length for Gaussianity, but the first adapts the i-vectors representation to a speaker recognition system based on LDA and two-covariance scoring when the second adapts it to a Gaussian-PLDA model. Significant performance improvements are demonstrated on the male and female telephone portion of NIST SRE 2010. Index Terms: i-vectors, probabilistic linear discriminant analysis, speaker recognition.