DeFT-AN: Dense Frequency-Time Attentive Network for Multichannel Speech Enhancement

In this study, we propose a dense frequency-time attentive network (DeFT-AN) for multichannel speech enhancement. DeFT-AN is a mask estimation network that predicts a complex spectral masking pattern for suppress-ing the noise and reverberation embedded in the short-time Fourier transform (STFT) of...

Full description

Saved in:
Bibliographic Details
Published in:IEEE signal processing letters 2023, Vol.30, p.1-5
Main Authors: Lee, Dongheon, Choi, Jung-Woo
Format: Article
Language:English
Subjects:
Audio signals
complex-spectral masking
Convolution
Convolutional neural networks
Fourier transforms
Intelligibility
multichannel
Noise measurement
Speech enhancement
Speech processing
Time-domain analysis
Time-frequency analysis
transformer
Transformers
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this study, we propose a dense frequency-time attentive network (DeFT-AN) for multichannel speech enhancement. DeFT-AN is a mask estimation network that predicts a complex spectral masking pattern for suppress-ing the noise and reverberation embedded in the short-time Fourier transform (STFT) of an input signal. The proposed mask estimation network incorporates three different types of blocksfor aggregatinginformationin thespatial, spectral, and temporal dimensions. It utilizes a spectral transformer with a modified feed-forward network and a temporal con-former with sequential dilated convolutions. The use of dense blocks and transformers dedicated to the three differ-ent characteristics of audio signals enables more compre-hensive enhancement in noisy and reverberant environ-ments. The remarkable performance of DeFT-AN over state-of-the-art multichannel models is demonstrated based on two popular noisy and reverberant datasets in terms of various metrics for speech quality and intelligibility.
ISSN:1070-9908
1558-2361
DOI:10.1109/LSP.2023.3244428