Generative x-vectors for text-independent speaker verification

TL;DR

This paper introduces generative x-vectors, a novel approach combining i-vector and x-vector information via a transformation model, significantly improving speaker verification performance especially for long-duration utterances.

Contribution

The paper proposes a new generative x-vector method using canonical correlation analysis to enhance speaker verification by integrating i-vector and x-vector features.

Findings

Generative x-vectors outperform baseline i-vector and x-vector systems.

The method improves performance on long-duration utterances.

Comparable results to fusion systems for short-duration utterances.

Abstract

Speaker verification (SV) systems using deep neural network embeddings, so-called the x-vector systems, are becoming popular due to its good performance superior to the i-vector systems. The fusion of these systems provides improved performance benefiting both from the discriminatively trained x-vectors and generative i-vectors capturing distinct speaker characteristics. In this paper, we propose a novel method to include the complementary information of i-vector and x-vector, that is called generative x-vector. The generative x-vector utilizes a transformation model learned from the i-vector and x-vector representations of the background data. Canonical correlation analysis is applied to derive this transformation model, which is later used to transform the standard x-vectors of the enrollment and test segments to the corresponding generative x-vectors. The SV experiments performed on the NIST SRE 2010 dataset demonstrate that the system using generative x-vectors provides considerably better performance than the baseline i-vector and x-vector systems. Furthermore, the generative x-vectors outperform the fusion of i-vector and x-vector systems for long-duration utterances, while yielding comparable results for short-duration utterances.