Multi-Level Transfer Learning from Near-Field to Far-Field Speaker Verification

TL;DR

This paper introduces a transfer learning approach for far-field speaker verification that employs feature-level and instance-level knowledge transfer within a teacher-student framework to improve domain-invariant embeddings, significantly reducing error rates.

Contribution

It proposes novel contrastive and pairwise distance transfer methods for domain adaptation in speaker verification, outperforming existing approaches.

Findings

13.9% reduction in EER on Full-eval trials

6.3% reduction in minDCF compared to winner's DenseNet

Performance close to fusion systems on challenging trials

Abstract

In far-field speaker verification, the performance of speaker embeddings is susceptible to degradation when there is a mismatch between the conditions of enrollment and test speech. To solve this problem, we propose the feature-level and instance-level transfer learning in the teacher-student framework to learn a domain-invariant embedding space. For the feature-level knowledge transfer, we develop the contrastive loss to transfer knowledge from teacher model to student model, which can not only decrease the intra-class distance, but also enlarge the inter-class distance. Moreover, we propose the instance-level pairwise distance transfer method to force the student model to preserve pairwise instances distance from the well optimized embedding space of the teacher model. On FFSVC 2020 evaluation set, our EER on Full-eval trials is relatively reduced by 13.9% compared with the fusion system result on Partial-eval trials of Task2. On Task1, compared with the winner's DenseNet result on Partial-eval trials, our minDCF on Full-eval trials is relatively reduced by 6.3%. On Task3, the EER and minDCF of our proposed method on Full-eval trials are very close to the result of the fusion system on Partial-eval trials. Our results also outperform other competitive domain adaptation methods.