Compact Speaker Embedding: lrx-vector

TL;DR

This paper introduces the lrx-vector, a low-rank factorized speaker embedding system that significantly reduces memory usage while maintaining recognition accuracy, by applying knowledge distillation and comparing it to SVD-based methods.

Contribution

The paper proposes the lrx-vector, a low-rank x-vector variant, and demonstrates its effectiveness in reducing model size without sacrificing performance.

Findings

Reduced model weights by 28% compared to full-rank x-vector.

Maintained recognition rate with 1.83% EER on VOiCES 2019 corpus.

Validated effectiveness of knowledge distillation in low-rank speaker embeddings.

Abstract

Deep neural networks (DNN) have recently been widely used in speaker recognition systems, achieving state-of-the-art performance on various benchmarks. The x-vector architecture is especially popular in this research community, due to its excellent performance and manageable computational complexity. In this paper, we present the lrx-vector system, which is the low-rank factorized version of the x-vector embedding network. The primary objective of this topology is to further reduce the memory requirement of the speaker recognition system. We discuss the deployment of knowledge distillation for training the lrx-vector system and compare against low-rank factorization with SVD. On the VOiCES 2019 far-field corpus we were able to reduce the weights by 28% compared to the full-rank x-vector system while keeping the recognition rate constant (1.83% EER).