Gaussian-Constrained training for speaker verification

TL;DR

This paper introduces a Gaussian-constrained training method for speaker verification neural models that removes the parametric classifier and enforces Gaussian distribution of speaker vectors, improving performance.

Contribution

It proposes a novel training approach that discards the classifier and constrains speaker vector distribution, addressing information leak and distribution issues in neural speaker verification models.

Findings

Improved speaker verification performance on VoxCeleb and SITW datasets.

Generated more representative and regular speaker embeddings.

Achieved consistent performance gains over traditional methods.

Abstract

Neural models, in particular the d-vector and x-vector architectures, have produced state-of-the-art performance on many speaker verification tasks. However, two potential problems of these neural models deserve more investigation. Firstly, both models suffer from `information leak', which means that some parameters participating in model training will be discarded during inference, i.e, the layers that are used as the classifier. Secondly, these models do not regulate the distribution of the derived speaker vectors. This `unconstrained distribution' may degrade the performance of the subsequent scoring component, e.g., PLDA. This paper proposes a Gaussian-constrained training approach that (1) discards the parametric classifier, and (2) enforces the distribution of the derived speaker vectors to be Gaussian. Our experiments on the VoxCeleb and SITW databases demonstrated that this new training approach produced more representative and regular speaker embeddings, leading to consistent performance improvement.