# IMSE: Efficient U-Net-based Speech Enhancement using Inception Depthwise Convolution and Amplitude-Aware Linear Attention

arXiv: https://arxiv.org/abs/2511.14515 · HTML: https://mlsanity.com/paper/2511.14515/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 17 of its 22 references that are in the map.
- First posted on arXiv 18 Nov 2025.

## Builds on (most cited first, 17 of 17)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [SEGAN: Speech Enhancement Generative Adversarial Network](https://mlsanity.com/paper/1703.09452/) (2017)
- [MetricGAN+: An Improved Version of MetricGAN for Speech Enhancement](https://mlsanity.com/paper/2104.03538/) (2021)
- [InceptionNeXt: When Inception Meets ConvNeXt](https://mlsanity.com/paper/2303.16900/) (2023)
- [MB-TaylorFormer: Multi-branch Efficient Transformer Expanded by Taylor Formula for Image Dehazing](https://mlsanity.com/paper/2308.14036/) (2023)
- [Demucs: Deep Extractor for Music Sources with extra unlabeled data remixed](https://mlsanity.com/paper/1909.01174/) (2019)
- [MP-SENet: A Speech Enhancement Model with Parallel Denoising of Magnitude and Phase Spectra](https://mlsanity.com/paper/2305.13686/) (2023)
- [PHASEN: A Phase-and-Harmonics-Aware Speech Enhancement Network](https://mlsanity.com/paper/1911.04697/) (2019)
- [TSTNN: Two-stage Transformer based Neural Network for Speech Enhancement in the Time Domain](https://mlsanity.com/paper/2103.09963/) (2021)
- [DPT-FSNet: Dual-path Transformer Based Full-band and Sub-band Fusion Network for Speech Enhancement](https://mlsanity.com/paper/2104.13002/) (2021)
- [MANNER: Multi-view Attention Network for Noise Erasure](https://mlsanity.com/paper/2203.02181/) (2022)
- [Dual-branch Attention-In-Attention Transformer for single-channel speech enhancement](https://mlsanity.com/paper/2110.06467/) (2021)
- [A Two-Stage U-Net for High-Fidelity Denoising of Historical Recordings](https://mlsanity.com/paper/2202.08702/) (2022)
- [Rectifying Magnitude Neglect in Linear Attention](https://mlsanity.com/paper/2507.00698/) (2025)
- [Multi-View Attention Transfer for Efficient Speech Enhancement](https://mlsanity.com/paper/2208.10367/) (2022)
- [MUSE: Flexible Voiceprint Receptive Fields and Multi-Path Fusion Enhanced Taylor Transformer for U-Net-based Speech Enhancement](https://mlsanity.com/paper/2406.04589/) (2024)

## Abstract (arXiv)

Achieving a balance between lightweight design and high performance remains a significant challenge for speech enhancement (SE) tasks on resource-constrained devices. Existing state-of-the-art methods, such as MUSE, have established a strong baseline with only 0.51M parameters by introducing a Multi-path Enhanced Taylor (MET) transformer and Deformable Embedding (DE). However, an in-depth analysis reveals that MUSE still suffers from efficiency bottlenecks: the MET module relies on a complex "approximate-compensate" mechanism to mitigate the limitations of Taylor-expansion-based attention, while the offset calculation for deformable embedding introduces additional computational burden. This paper proposes IMSE, a systematically optimized and ultra-lightweight network. We introduce two core innovations: 1) Replacing the MET module with Amplitude-Aware Linear Attention (MALA). MALA fundamentally rectifies the "amplitude-ignoring" problem in linear attention by explicitly preserving the norm information of query vectors in the attention calculation, achieving efficient global modeling without an auxiliary compensation branch. 2) Replacing the DE module with Inception Depthwise Convolution (IDConv). IDConv borrows the Inception concept, decomposing large-kernel operations into efficient parallel branches (square, horizontal, and vertical strips), thereby capturing spectrogram features with extremely low parameter redundancy. Extensive experiments on the VoiceBank+DEMAND dataset demonstrate that, compared to the MUSE baseline, IMSE significantly reduces the parameter count by 16.8% (from 0.513M to 0.427M) while achieving competitive performance comparable to the state-of-the-art on the PESQ metric (3.373). This study sets a new benchmark for the trade-off between model size and speech quality in ultra-lightweight speech enhancement.