Self-supervised Speaker Recognition Training Using Human-Machine Dialogues

TL;DR

This paper introduces a self-supervised training approach for speaker recognition using noisy human-machine dialogues, employing a rejection mechanism to improve model pretraining and significantly reduce error rates.

Contribution

It proposes a novel rejection mechanism to select acoustically homogeneous dialogues for self-supervised pretraining in speaker recognition.

Findings

Achieved 27.10% EER reduction with dialogue pretraining and rejection mechanism.

Compared reconstruction-based and contrastive-learning-based methods, favoring the latter.

Demonstrated significant performance improvements over previous methods.

Abstract

Speaker recognition, recognizing speaker identities based on voice alone, enables important downstream applications, such as personalization and authentication. Learning speaker representations, in the context of supervised learning, heavily depends on both clean and sufficient labeled data, which is always difficult to acquire. Noisy unlabeled data, on the other hand, also provides valuable information that can be exploited using self-supervised training methods. In this work, we investigate how to pretrain speaker recognition models by leveraging dialogues between customers and smart-speaker devices. However, the supervisory information in such dialogues is inherently noisy, as multiple speakers may speak to a device in the course of the same dialogue. To address this issue, we propose an effective rejection mechanism that selectively learns from dialogues based on their acoustic homogeneity. Both reconstruction-based and contrastive-learning-based self-supervised methods are compared. Experiments demonstrate that the proposed method provides significant performance improvements, superior to earlier work. Dialogue pretraining when combined with the rejection mechanism yields 27.10% equal error rate (EER) reduction in speaker recognition, compared to a model without self-supervised pretraining.