DBNet: A Dual-branch Network Architecture Processing on Spectrum and Waveform for Single-channel Speech Enhancement

被引:5
|
作者
Zhang, Kanghao [1 ]
He, Shulin [1 ]
Li, Hao [1 ]
Zhang, Xueliang [1 ]
机构
[1] Inner Mongolia Univ, Coll Comp Sci, Hohhot, Peoples R China
来源
INTERSPEECH 2021 | 2021年
关键词
Speech enhancement; time domain; frequency domain; real-time; NEURAL-NETWORK; CNN;
D O I
10.21437/Interspeech.2021-1042
中图分类号
R36 [病理学]; R76 [耳鼻咽喉科学];
学科分类号
100104 ; 100213 ;
摘要
In real acoustic environment, speech enhancement is an arduous task to improve the quality and intelligibility of speech interfered by background noise and reverberation. Over the past years, deep learning has shown great potential on speech enhancement. In this paper, we propose a novel real-time framework called DBNet which is a dual-branch structure with alternate interconnection. Each branch incorporates an encoderdecoder architecture with skip connections. The two branches are responsible for spectrum and waveform modeling, respectively. A bridge layer is adopted to exchange information between the two branches. Systematic evaluation and comparison show that the proposed system substantially outperforms related algorithms under very challenging environments. And in INTERSPEECH 2021 Deep Noise Suppression (DNS) challenge, the proposed system ranks the top 8 in real-time track 1 in terms of the Mean Opinion Score (MOS) of the ITU-T P.835 framework.
引用
收藏
页码:2821 / 2825
页数:5
相关论文
共 50 条
  • [21] PhaseDCN: A Phase-Enhanced Dual-Path Dilated Convolutional Network for Single-Channel Speech Enhancement
    Zhang, Lu
    Wang, Mingjiang
    Zhang, Qiquan
    Wang, Xinsheng
    Liu, Ming
    IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING, 2021, 29 : 2561 - 2574
  • [22] GAUSSIAN DENSITY GUIDED DEEP NEURAL NETWORK FOR SINGLE-CHANNEL SPEECH ENHANCEMENT
    Chai, Li
    Du, Jun
    Wang, Yan-nan
    2017 IEEE 27TH INTERNATIONAL WORKSHOP ON MACHINE LEARNING FOR SIGNAL PROCESSING, 2017,
  • [23] DDP-Unet: A mapping neural network for single-channel speech enhancement
    Chen, Haoxiang
    Xu, Yanyan
    Ke, Dengfeng
    Su, Kaile
    COMPUTER SPEECH AND LANGUAGE, 2025, 93
  • [24] TFDense-GAN: a generative adversarial network for single-channel speech enhancement
    Chen, Haoxiang
    Zhang, Jinxiu
    Fu, Yaogang
    Zhou, Xintong
    Wang, Ruilong
    Xu, Yanyan
    Ke, Dengfeng
    EURASIP JOURNAL ON ADVANCES IN SIGNAL PROCESSING, 2025, 2025 (01):
  • [25] NOISE-ADAPTIVE DEEP NEURAL NETWORK FOR SINGLE-CHANNEL SPEECH ENHANCEMENT
    Chung, Hanwook
    Kim, Taesup
    Plourde, Eric
    Champagne, Benoit
    2018 IEEE 28TH INTERNATIONAL WORKSHOP ON MACHINE LEARNING FOR SIGNAL PROCESSING (MLSP), 2018,
  • [26] Single-channel Speech Enhancement Student under Multi-channel Speech Enhancement Teacher
    Zhang, Yuzhu
    Zhang, Hui
    Zhang, Xueliang
    PROCEEDINGS OF 2022 ASIA-PACIFIC SIGNAL AND INFORMATION PROCESSING ASSOCIATION ANNUAL SUMMIT AND CONFERENCE (APSIPA ASC), 2022, : 372 - 377
  • [27] QDPN - Quasi-dual-path Network for single-channel Speech Separation
    Rixen, Joel
    Renz, Matthias
    INTERSPEECH 2022, 2022, : 5353 - 5357
  • [28] Single-Channel Speech Enhancement Based on Psychoacoustic Masking
    Zhou, Tingting
    Zeng, Yumin
    Wang, Rongrong
    JOURNAL OF THE AUDIO ENGINEERING SOCIETY, 2017, 65 (04): : 272 - 284
  • [29] Single-channel speech enhancement by subspace affinity minimization
    Tran, Dung N.
    Koishida, Kazuhito
    INTERSPEECH 2020, 2020, : 2447 - 2451
  • [30] Single-channel speech enhancement using colored spectrograms
    Gul, Sania
    Khan, Muhammad Salman
    Fazeel, Muhammad
    COMPUTER SPEECH AND LANGUAGE, 2024, 86