Saved in:
Bibliographic Details
Main Authors: Zhao, Siyi, Wang, Wei, Qian, Yanmin
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.21833
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915515899838464
author Zhao, Siyi
Wang, Wei
Qian, Yanmin
author_facet Zhao, Siyi
Wang, Wei
Qian, Yanmin
contents Recent advancements in automatic speech recognition (ASR) have achieved notable progress, whereas robustness in noisy environments remains challenging. While speech enhancement (SE) front-ends are widely used to mitigate noise as a preprocessing step for ASR, they often introduce computational non-negligible overhead. This paper proposes optimizations to reduce SE computational costs without compromising ASR performance. Our approach integrates layer-wise frame resampling and progressive sub-band pruning. Frame resampling downsamples inputs within layers, utilizing residual connections to mitigate information loss. Simultaneously, sub-band pruning progressively excludes less informative frequency bands, further reducing computational demands. Extensive experiments on synthetic and real-world noisy datasets demonstrate that our system reduces SE computational overhead over 66 compared to the standard BSRNN, while maintaining strong ASR performance.
format Preprint
id arxiv_https___arxiv_org_abs_2509_21833
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Lightweight Front-end Enhancement for Robust ASR via Frame Resampling and Sub-Band Pruning
Zhao, Siyi
Wang, Wei
Qian, Yanmin
Sound
Recent advancements in automatic speech recognition (ASR) have achieved notable progress, whereas robustness in noisy environments remains challenging. While speech enhancement (SE) front-ends are widely used to mitigate noise as a preprocessing step for ASR, they often introduce computational non-negligible overhead. This paper proposes optimizations to reduce SE computational costs without compromising ASR performance. Our approach integrates layer-wise frame resampling and progressive sub-band pruning. Frame resampling downsamples inputs within layers, utilizing residual connections to mitigate information loss. Simultaneously, sub-band pruning progressively excludes less informative frequency bands, further reducing computational demands. Extensive experiments on synthetic and real-world noisy datasets demonstrate that our system reduces SE computational overhead over 66 compared to the standard BSRNN, while maintaining strong ASR performance.
title Lightweight Front-end Enhancement for Robust ASR via Frame Resampling and Sub-Band Pruning
topic Sound
url https://arxiv.org/abs/2509.21833