Self-Supervised Speaker Verification Using Dynamic Loss-Gate and Label Correction

TL;DR

This paper introduces a novel self-supervised speaker verification method that uses dynamic loss-gate and label correction to handle unreliable pseudo labels, resulting in faster convergence and significant performance improvements.

Contribution

The paper proposes DLG-LC, a new approach combining Gaussian Mixture Model-based loss gating and label correction, enhancing self-supervised speaker verification performance.

Findings

Achieves up to 18.35% relative improvement on Voxceleb1 trials.

Converges faster than previous self-supervised methods.

Effectively utilizes unreliable data through label correction.

Abstract

For self-supervised speaker verification, the quality of pseudo labels decides the upper bound of the system due to the massive unreliable labels. In this work, we propose dynamic loss-gate and label correction (DLG-LC) to alleviate the performance degradation caused by unreliable estimated labels. In DLG, we adopt Gaussian Mixture Model (GMM) to dynamically model the loss distribution and use the estimated GMM to distinguish the reliable and unreliable labels automatically. Besides, to better utilize the unreliable data instead of dropping them directly, we correct the unreliable label with model predictions. Moreover, we apply the negative-pairs-free DINO framework in our experiments for further improvement. Compared to the best-known speaker verification system with self-supervised learning, our proposed DLG-LC converges faster and achieves 11.45%, 18.35% and 15.16% relative improvement on Vox-O, Vox-E and Vox-H trials of Voxceleb1 evaluation dataset.