Self-Supervised Training of Speaker Encoder with Multi-Modal Diverse Positive Pairs

TL;DR

This paper introduces a multi-modal contrastive learning approach for speaker recognition that leverages speech and face data to create diverse positive pairs, significantly improving self-supervised speaker encoder performance.

Contribution

It proposes a novel multi-modal sampling strategy for contrastive learning, enhancing the robustness of self-supervised speaker encoders without using speaker labels.

Findings

Achieved state-of-the-art EER on VoxCeleb1 with self-supervised training.

Outperformed existing self-supervised methods by a large margin.

Demonstrated robustness across multiple datasets.

Abstract

We study a novel neural architecture and its training strategies of speaker encoder for speaker recognition without using any identity labels. The speaker encoder is trained to extract a fixed-size speaker embedding from a spoken utterance of various length. Contrastive learning is a typical self-supervised learning technique. However, the quality of the speaker encoder depends very much on the sampling strategy of positive and negative pairs. It is common that we sample a positive pair of segments from the same utterance. Unfortunately, such poor-man's positive pairs (PPP) lack necessary diversity for the training of a robust encoder. In this work, we propose a multi-modal contrastive learning technique with novel sampling strategies. By cross-referencing between speech and face data, we study a method that finds diverse positive pairs (DPP) for contrastive learning, thus improving the robustness of the speaker encoder. We train the speaker encoder on the VoxCeleb2 dataset without any speaker labels, and achieve an equal error rate (EER) of 2.89\%, 3.17\% and 6.27\% under the proposed progressive clustering strategy, and an EER of 1.44\%, 1.77\% and 3.27\% under the two-stage learning strategy with pseudo labels, on the three test sets of VoxCeleb1. This novel solution outperforms the state-of-the-art self-supervised learning methods by a large margin, at the same time, achieves comparable results with the supervised learning counterpart. We also evaluate our self-supervised learning technique on LRS2 and LRW datasets, where the speaker information is unknown. All experiments suggest that the proposed neural architecture and sampling strategies are robust across datasets.