LORT: Locally Refined Convolution and Taylor Transformer for Monaural Speech Enhancement

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Junyu, Lin, Zizhen, Wang, Tianrui, Ge, Meng, Wang, Longbiao, Dang, Jianwu
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908563422576640
author Wang, Junyu
Lin, Zizhen
Wang, Tianrui
Ge, Meng
Wang, Longbiao
Dang, Jianwu
author_facet Wang, Junyu
Lin, Zizhen
Wang, Tianrui
Ge, Meng
Wang, Longbiao
Dang, Jianwu
contents Achieving superior enhancement performance while maintaining a low parameter count and computational complexity remains a challenge in the field of speech enhancement. In this paper, we introduce LORT, a novel architecture that integrates spatial-channel enhanced Taylor Transformer and locally refined convolution for efficient and robust speech enhancement. We propose a Taylor multi-head self-attention (T-MSA) module enhanced with spatial-channel enhancement attention (SCEA), designed to facilitate inter-channel information exchange and alleviate the spatial attention limitations inherent in Taylor-based Transformers. To complement global modeling, we further present a locally refined convolution (LRC) block that integrates convolutional feed-forward layers, time-frequency dense local convolutions, and gated units to capture fine-grained local details. Built upon a U-Net-like encoder-decoder structure with only 16 output channels in the encoder, LORT processes noisy inputs through multi-resolution T-MSA modules using alternating downsampling and upsampling operations. The enhanced magnitude and phase spectra are decoded independently and optimized through a composite loss function that jointly considers magnitude, complex, phase, discriminator, and consistency objectives. Experimental results on the VCTK+DEMAND and DNS Challenge datasets demonstrate that LORT achieves competitive or superior performance to state-of-the-art (SOTA) models with only 0.96M parameters, highlighting its effectiveness for real-world speech enhancement applications with limited computational resources.
format Preprint
id arxiv_https___arxiv_org_abs_2509_23832
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LORT: Locally Refined Convolution and Taylor Transformer for Monaural Speech Enhancement
Wang, Junyu
Lin, Zizhen
Wang, Tianrui
Ge, Meng
Wang, Longbiao
Dang, Jianwu
Audio and Speech Processing
Sound
Achieving superior enhancement performance while maintaining a low parameter count and computational complexity remains a challenge in the field of speech enhancement. In this paper, we introduce LORT, a novel architecture that integrates spatial-channel enhanced Taylor Transformer and locally refined convolution for efficient and robust speech enhancement. We propose a Taylor multi-head self-attention (T-MSA) module enhanced with spatial-channel enhancement attention (SCEA), designed to facilitate inter-channel information exchange and alleviate the spatial attention limitations inherent in Taylor-based Transformers. To complement global modeling, we further present a locally refined convolution (LRC) block that integrates convolutional feed-forward layers, time-frequency dense local convolutions, and gated units to capture fine-grained local details. Built upon a U-Net-like encoder-decoder structure with only 16 output channels in the encoder, LORT processes noisy inputs through multi-resolution T-MSA modules using alternating downsampling and upsampling operations. The enhanced magnitude and phase spectra are decoded independently and optimized through a composite loss function that jointly considers magnitude, complex, phase, discriminator, and consistency objectives. Experimental results on the VCTK+DEMAND and DNS Challenge datasets demonstrate that LORT achieves competitive or superior performance to state-of-the-art (SOTA) models with only 0.96M parameters, highlighting its effectiveness for real-world speech enhancement applications with limited computational resources.
title LORT: Locally Refined Convolution and Taylor Transformer for Monaural Speech Enhancement
topic Audio and Speech Processing
Sound
url https://arxiv.org/abs/2509.23832