Enhancing Age-Related Robustness in Children Speaker Verification

TL;DR

This paper proposes a Feature Transform Adapter and Synthetic Audio Augmentation to improve age-related robustness in children's speaker verification, demonstrating significant error rate reductions over multiple years.

Contribution

Introduction of a novel Feature Transform Adapter and Synthetic Audio Augmentation to enhance age-invariance in children's speaker verification systems.

Findings

Reduced equal error rate by up to 19.4% over three-year gaps.

Created a new longitudinal dataset for evaluating age-related robustness.

Demonstrated improved inter-year verification performance.

Abstract

One of the main challenges in children's speaker verification (C-SV) is the significant change in children's voices as they grow. In this paper, we propose two approaches to improve age-related robustness in C-SV. We first introduce a Feature Transform Adapter (FTA) module that integrates local patterns into higher-level global representations, reducing overfitting to specific local features and improving the inter-year SV performance of the system. We then employ Synthetic Audio Augmentation (SAA) to increase data diversity and size, thereby improving robustness against age-related changes. Since the lack of longitudinal speech datasets makes it difficult to measure age-related robustness of C-SV systems, we introduce a longitudinal dataset to assess inter-year verification robustness of C-SV systems. By integrating both of our proposed methods, the average equal error rate was reduced by 19.4%, 13.0%, and 6.1% in the one-year, two-year, and three-year gap inter-year evaluation sets, respectively, compared to the baseline.