Research

Paper

TESTING March 09, 2026

Language-Invariant Multilingual Speaker Verification for the TidyVoice 2026 Challenge

Authors

Ze Li, Xiaoxiao Miao, Juan Liu, Ming Li

Abstract

Multilingual speaker verification (SV) remains challenging due to limited cross-lingual data and language-dependent information in speaker embeddings. This paper presents a language-invariant multilingual SV system for the TidyVoice 2026 Challenge. We adopt the multilingual self-supervised w2v-BERT 2.0 model as the backbone, enhanced with Layer Adapters and Multi-scale Feature Aggregation to better exploit multi-layer representations. A language-adversarial training strategy with a Gradient Reversal Layer is applied to promote language-invariant speaker embeddings. Moreover, a multilingual zero-shot text-to-speech system is used to synthesize speech in multiple languages, improving language diversity. Experimental results demonstrate that fine-tuning the large-scale pretrained model yields competitive performance, while language-adversarial training further enhances robustness. In addition, synthetic speech augmentation provides additional gains under limited training data conditions. Source code is available at https://github.com/ZXHY-82/LI-MSV-TidyVoice2026.

Metadata

arXiv ID: 2603.08092
Provider: ARXIV
Primary Category: eess.AS
Published: 2026-03-09
Fetched: 2026-03-10 05:43

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.08092v1</id>\n    <title>Language-Invariant Multilingual Speaker Verification for the TidyVoice 2026 Challenge</title>\n    <updated>2026-03-09T08:33:13Z</updated>\n    <link href='https://arxiv.org/abs/2603.08092v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.08092v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Multilingual speaker verification (SV) remains challenging due to limited cross-lingual data and language-dependent information in speaker embeddings. This paper presents a language-invariant multilingual SV system for the TidyVoice 2026 Challenge. We adopt the multilingual self-supervised w2v-BERT 2.0 model as the backbone, enhanced with Layer Adapters and Multi-scale Feature Aggregation to better exploit multi-layer representations. A language-adversarial training strategy with a Gradient Reversal Layer is applied to promote language-invariant speaker embeddings. Moreover, a multilingual zero-shot text-to-speech system is used to synthesize speech in multiple languages, improving language diversity. Experimental results demonstrate that fine-tuning the large-scale pretrained model yields competitive performance, while language-adversarial training further enhances robustness. In addition, synthetic speech augmentation provides additional gains under limited training data conditions. Source code is available at https://github.com/ZXHY-82/LI-MSV-TidyVoice2026.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='eess.AS'/>\n    <published>2026-03-09T08:33:13Z</published>\n    <arxiv:comment>submitted to Interspeech 2026</arxiv:comment>\n    <arxiv:primary_category term='eess.AS'/>\n    <author>\n      <name>Ze Li</name>\n    </author>\n    <author>\n      <name>Xiaoxiao Miao</name>\n    </author>\n    <author>\n      <name>Juan Liu</name>\n    </author>\n    <author>\n      <name>Ming Li</name>\n    </author>\n  </entry>"
}