Unified Architecture and Unsupervised Speech Disentanglement for Speaker Embedding-Free Enrollment in Personalized Speech Enhancement

TL;DR

This paper introduces unified models for speech enhancement and personalized speech enhancement that incorporate unsupervised speech disentanglement, improving robustness and performance across varying enrollment speech conditions.

Contribution

The paper proposes two novel models, USEF-PNet and DSEF-PNet, unifying SE and PSE tasks and employing unsupervised speech disentanglement to enhance robustness against enrollment speech variations.

Findings

Models outperform previous methods on Libri2Mix and VoiceBank DEMAND datasets.

Unsupervised speech disentanglement reduces interference from emotion and content.

Long-short enrollment pairing improves performance regardless of enrollment duration.

Abstract

Conventional speech enhancement (SE) aims to improve speech perception and intelligibility by suppressing noise without requiring enrollment speech as reference, whereas personalized SE (PSE) addresses the cocktail party problem by extracting a target speaker's speech using enrollment speech. While these two tasks tackle different yet complementary challenges in speech signal processing, they often share similar model architectures, with PSE incorporating an additional branch to process enrollment speech. This suggests developing a unified model capable of efficiently handling both SE and PSE tasks, thereby simplifying deployment while maintaining high performance. However, PSE performance is sensitive to variations in enrollment speech, like emotional tone, which limits robustness in real-world applications. To address these challenges, we propose two novel models, USEF-PNet and DSEF-PNet, both extending our previous SEF-PNet framework. USEF-PNet introduces a unified architecture for processing enrollment speech, integrating SE and PSE into a single framework to enhance performance and streamline deployment. Meanwhile, DSEF-PNet incorporates an unsupervised speech disentanglement approach by pairing a mixture speech with two different enrollment utterances and enforcing consistency in the extracted target speech. This strategy effectively isolates high-quality speaker identity information from enrollment speech, reducing interference from factors such as emotion and content, thereby improving PSE robustness. Additionally, we explore a long-short enrollment pairing (LSEP) strategy to examine the impact of enrollment speech duration during both training and evaluation. Extensive experiments on the Libri2Mix and VoiceBank DEMAND demonstrate that our proposed USEF-PNet, DSEF-PNet all achieve substantial performance improvements, with random enrollment duration performing slightly better.