Two-Step Band-Split Neural Network Approach for Full-Band Residual Echo Suppression

Zihan Zhang; Shimin Zhang; Mingshuai Liu; Yanhong Leng; Zhe Han; Li Chen; Lei Xie

doi:10.1109/ICASSP49357.2023.10096953

Two-Step Band-Split Neural Network Approach for Full-Band Residual Echo Suppression

Zihan Zhang, Shimin Zhang, Mingshuai Liu, Yanhong Leng, Zhe Han, Li Chen, Lei Xie

School of Computer Science

Research output: Contribution to journal › Conference article › peer-review

6 Scopus citations

Abstract

This paper describes a Two-step Band-split Neural Network (TBNN) approach for full-band acoustic echo cancellation. Specifically, after linear filtering, we split the full-band signal into wideband (16KHz) and high-band (16-48KHz) for residual echo removal with lower modeling difficulty. The wide-band signal is processed by an updated gated convolutional recurrent network (GCRN) with U² encoder while the high-band signal is processed by a high-band post-filter net with lower complexity. Our approach submitted to ICASSP 2023 AEC Challenge has achieved an overall mean opinion score (MOS) of 4.344 and a word accuracy (WAcc) ratio of 0.795, leading to the 2^nd (tied) in the ranking of the non-personalized track.

Original language	English
Journal	ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
DOIs	https://doi.org/10.1109/ICASSP49357.2023.10096953
State	Published - 2023
Event	48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023 - Rhodes Island, Greece Duration: 4 Jun 2023 → 10 Jun 2023

Keywords

Acoustic echo cancellation
band-split
noise suppression
two-step network

Access to Document

10.1109/ICASSP49357.2023.10096953

Cite this

@article{557954937a63466d9ab4e41f539371d9,

title = "Two-Step Band-Split Neural Network Approach for Full-Band Residual Echo Suppression",

abstract = "This paper describes a Two-step Band-split Neural Network (TBNN) approach for full-band acoustic echo cancellation. Specifically, after linear filtering, we split the full-band signal into wideband (16KHz) and high-band (16-48KHz) for residual echo removal with lower modeling difficulty. The wide-band signal is processed by an updated gated convolutional recurrent network (GCRN) with U2 encoder while the high-band signal is processed by a high-band post-filter net with lower complexity. Our approach submitted to ICASSP 2023 AEC Challenge has achieved an overall mean opinion score (MOS) of 4.344 and a word accuracy (WAcc) ratio of 0.795, leading to the 2nd (tied) in the ranking of the non-personalized track.",

keywords = "Acoustic echo cancellation, band-split, noise suppression, two-step network",

author = "Zihan Zhang and Shimin Zhang and Mingshuai Liu and Yanhong Leng and Zhe Han and Li Chen and Lei Xie",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023 ; Conference date: 04-06-2023 Through 10-06-2023",

year = "2023",

doi = "10.1109/ICASSP49357.2023.10096953",

language = "英语",

journal = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",

issn = "1520-6149",

}

TY - JOUR

T1 - Two-Step Band-Split Neural Network Approach for Full-Band Residual Echo Suppression

AU - Zhang, Zihan

AU - Zhang, Shimin

AU - Liu, Mingshuai

AU - Leng, Yanhong

AU - Han, Zhe

AU - Chen, Li

AU - Xie, Lei

PY - 2023

Y1 - 2023

N2 - This paper describes a Two-step Band-split Neural Network (TBNN) approach for full-band acoustic echo cancellation. Specifically, after linear filtering, we split the full-band signal into wideband (16KHz) and high-band (16-48KHz) for residual echo removal with lower modeling difficulty. The wide-band signal is processed by an updated gated convolutional recurrent network (GCRN) with U2 encoder while the high-band signal is processed by a high-band post-filter net with lower complexity. Our approach submitted to ICASSP 2023 AEC Challenge has achieved an overall mean opinion score (MOS) of 4.344 and a word accuracy (WAcc) ratio of 0.795, leading to the 2nd (tied) in the ranking of the non-personalized track.

AB - This paper describes a Two-step Band-split Neural Network (TBNN) approach for full-band acoustic echo cancellation. Specifically, after linear filtering, we split the full-band signal into wideband (16KHz) and high-band (16-48KHz) for residual echo removal with lower modeling difficulty. The wide-band signal is processed by an updated gated convolutional recurrent network (GCRN) with U2 encoder while the high-band signal is processed by a high-band post-filter net with lower complexity. Our approach submitted to ICASSP 2023 AEC Challenge has achieved an overall mean opinion score (MOS) of 4.344 and a word accuracy (WAcc) ratio of 0.795, leading to the 2nd (tied) in the ranking of the non-personalized track.

KW - Acoustic echo cancellation

KW - band-split

KW - noise suppression

KW - two-step network

UR - http://www.scopus.com/inward/record.url?scp=85174055739&partnerID=8YFLogxK

U2 - 10.1109/ICASSP49357.2023.10096953

DO - 10.1109/ICASSP49357.2023.10096953

M3 - 会议文章

AN - SCOPUS:85174055739

SN - 1520-6149

JO - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

JF - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

T2 - 48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023

Y2 - 4 June 2023 through 10 June 2023

ER -

Two-Step Band-Split Neural Network Approach for Full-Band Residual Echo Suppression

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this