CONTRASTIVE SPEAKER EMBEDDING WITH SEQUENTIAL DISENTANGLEMENT

Youzhi Tu, Man Wai Mak, Jen Tzung Chien

研究成果: Conference contribution同行評審

3 引文 斯高帕斯(Scopus)

摘要

Contrastive speaker embedding assumes that the contrast between the positive and negative pairs of speech segments is attributed to speaker identity only. However, this assumption is incorrect because speech signals contain not only speaker identity but also linguistic content. In this paper, we propose a contrastive learning framework with sequential disentanglement to remove linguistic content by incorporating a disentangled sequential variational autoencoder (DSVAE) into the conventional SimCLR framework. The DSVAE aims to disentangle speaker factors from content factors in an embedding space so that only the speaker factors are used for constructing a contrastive loss objective. Because content factors have been removed from the contrastive learning, the resulting speaker embeddings will be content-invariant. Experimental results on VoxCeleb1-test show that the proposed method consistently outperforms SimCLR. This suggests that applying sequential disentanglement is beneficial to learning speaker-discriminative embeddings.

原文English
主出版物標題2024 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2024 - Proceedings
發行者Institute of Electrical and Electronics Engineers Inc.
頁面10891-10895
頁數5
ISBN(電子)9798350344851
DOIs
出版狀態Published - 2024
事件49th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2024 - Seoul, 韓國
持續時間: 14 4月 202419 4月 2024

出版系列

名字ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
ISSN(列印)1520-6149

Conference

Conference49th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2024
國家/地區韓國
城市Seoul
期間14/04/2419/04/24

指紋

深入研究「CONTRASTIVE SPEAKER EMBEDDING WITH SEQUENTIAL DISENTANGLEMENT」主題。共同形成了獨特的指紋。

引用此