Papers

Real-Time Neural Speech Enhancement Based on Temporal Refinement Network and Channel-Wise Gating Methods

International Journal
2021~
작성자
dsp
작성일
2023-01-30 16:43
조회
2361
Authors : Jinyoung Lee, Hong-Goo Kang

Year : 2023

Publisher / Conference : Digital Signal Processing

Volume : 133

Research area : Speech Signal Processing, Speech Enhancement

Presentation/Publication date : 08 December 2022

Presentation : None

Neural speech enhancement systems have seen dramatic improvements in performance recently. However, it is still difficult to create systems that can operate in real-time, with low delay, low complexity, and causality. In this paper, we propose a temporal and channel attention framework for a U-Net-based speech enhancement architecture that uses short analysis frame lengths. Specifically, we propose an attention-based temporal refinement network (TRN) that estimates convolutional features subject to the importance of temporal location. By adding the TRN output to the channel-attentive convolution output, we can further enhance speech-related features even in low-attentive channel outputs. To further improve the representation power of the convolutional features, we also apply a squeeze-and-excitation (SE)-based channel attention mechanism for three different network modules: main convolutional blocks after processing the TRN, skip connections, and residual connections in the bottleneck recurrent neural network (RNN) layer. In particular, a channel-wise gate architecture placed on the skip connections and residual connections reliably controls the data flow, which avoids transferring redundant information to the following stages. We show the effectiveness of the proposed TRN and channel-wise gating methods by visualizing the spectral characteristics of the corresponding features, evaluating overall enhancement performance, and performing ablation studies in various configurations. Our proposed real-time enhancement system outperforms several recent neural enhancement models in terms of quality, model size, and complexity.
전체 368
27 International Conference Seyun Um, Jihyun Kim, Jihyun Lee, Hong-Goo Kang "Facetron: A Multi-speaker Face-to-Speech Model based on Cross-Modal Latent Representations" in EUSIPCO, 2023
26 International Conference Hejung Yang, Hong-Goo Kang "Feature Normalization for Fine-tuning Self-Supervised Models in Speech Enhancement" in INTERSPEECH, 2023
25 International Conference Jihyun Kim, Hong-Goo Kang "Contrastive Learning based Deep Latent Masking for Music Source Seperation" in INTERSPEECH, 2023
24 International Conference Woo-Jin Chung, Doyeon Kim, Soo-Whan Chung, Hong-Goo Kang "MF-PAM: Accurate Pitch Estimation through Periodicity Analysis and Multi-level Feature Fusion" in INTERSPEECH, 2023
23 International Conference Hyungchan Yoon, Seyun Um, Changhwan Kim, Hong-Goo Kang "Adversarial Learning of Intermediate Acoustic Feature for End-to-End Lightweight Text-to-Speech" in INTERSPEECH, 2023
22 International Conference Hyungchan Yoon, Changhwan Kim, Eunwoo Song, Hyun-Wook Yoon, Hong-Goo Kang "Pruning Self-Attention for Zero-Shot Multi-Speaker Text-to-Speech" in INTERSPEECH, 2023
21 International Conference Doyeon Kim, Soo-Whan Chung, Hyewon Han, Youna Ji, Hong-Goo Kang "HD-DEMUCS: General Speech Restoration with Heterogeneous Decoders" in INTERSPEECH, 2023
20 Domestic Conference Jihyun Lee, Wootaek Lim, Hong-Goo Kang "음성 압축에서의 심층 신경망 기반 장구간 예측" in 한국방송·미디어공학회 2023년 하계학술대회, 2023
19 Domestic Conference Hwayeon Kim, Hong-Goo Kang "Band-Split based Dual-Path Convolution Recurrent Network for Music Source Separation" in 2023년도 한국음향학회 춘계학술발표대회 및 제38회 수중음향학 학술발표회, 2023
18 International Conference Zhenyu Piao, Miseul Kim, Hyungchan Yoon, Hong-Goo Kang "HappyQuokka System for ICASSP 2023 Auditory EEG Challenge" in ICASSP, 2023