Voice activity detection using a local-global attention model. (30th June 2022)

• Record Type: Journal Article
• Authors: Li, Shu; Li, Ye; Feng, Tao; Shi, Jinze; Zhang, Peng
• Is Part Of: Applied acoustics. Volume 195(2022)
• Publication Date: 2022-06-30
• Subjects: Voice activity detection -- Long short-term memory network -- Attention mechanism -- Deep learning; Acoustical engineering -- Periodicals; Periodicals; 620.2
• DOI: 10.1016/j.apacoust.2022.108802
• Languages: English
• ISSNs: 0003-682X

Abstract

Highlights: Local attention is executed in attention-enhanced LSTMs for local context modeling. Global attention ranks global contextual information. Both local attention and global attention are based on multi-head attention. Relative position encoding is applied to global attention. The loss function for attention score is designed to adjust attention distribution. Abstract: Voice activity detection (VAD) is an essential initial step of speech signal processing, and greatly affects the timeliness and accuracy of the system. Although many novel methods have been proposed to promote the performance of VAD under low signal-to-noise ratios (SNRs), the robustness to very low SNRs and unknown noisy environments has yet to be enhanced. In this paper, we propose a fusion model of local attention and global attention to strengthen the attention mechanism currently applied in VAD methods. First, based on self-attention, the local attention cooperates with long short-term memory networks (LSTMs) to achieve efficient use of local contextual information; and then the global attention measures global contextual information to focus on the most appropriate area of contextual frames. The experimental results show that compared with the state-of-the-art VAD methods, the proposed approach achieves better performance under low SNRs such as −15 dB and non-stationary noisy conditions.