Saved in:
Bibliographic Details
Main Authors: Hannan, Abdul, Falavigna, Daniele, Brutti, Alessio
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2507.07954
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915382321741824
author Hannan, Abdul
Falavigna, Daniele
Brutti, Alessio
author_facet Hannan, Abdul
Falavigna, Daniele
Brutti, Alessio
contents Curating foundation speech models for edge and IoT settings, where computational resources vary over time, requires dynamic architectures featuring adaptable reduction strategies. One emerging approach is layer dropping ($\mathcal{LD}$) which skips fraction of the layers of a backbone network during inference to reduce the computational load. This allows transforming static models into dynamic ones. However, existing approaches exhibit limitations either in the mode of selecting layers or by significantly modifying the neural architecture. To this end, we propose input-driven $\mathcal{LD}$ that employs the network's input features and a lightweight layer selecting network to determine the optimum combination of processing layers. Extensive experimentation on 4 speech and audio public benchmarks, using two different pre-trained foundation models, demonstrates the effectiveness of our approach, thoroughly outperforming random dropping and producing on-par (or better) results to early exit.
format Preprint
id arxiv_https___arxiv_org_abs_2507_07954
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Input Conditioned Layer Dropping in Speech Foundation Models
Hannan, Abdul
Falavigna, Daniele
Brutti, Alessio
Sound
Computer Vision and Pattern Recognition
Audio and Speech Processing
Curating foundation speech models for edge and IoT settings, where computational resources vary over time, requires dynamic architectures featuring adaptable reduction strategies. One emerging approach is layer dropping ($\mathcal{LD}$) which skips fraction of the layers of a backbone network during inference to reduce the computational load. This allows transforming static models into dynamic ones. However, existing approaches exhibit limitations either in the mode of selecting layers or by significantly modifying the neural architecture. To this end, we propose input-driven $\mathcal{LD}$ that employs the network's input features and a lightweight layer selecting network to determine the optimum combination of processing layers. Extensive experimentation on 4 speech and audio public benchmarks, using two different pre-trained foundation models, demonstrates the effectiveness of our approach, thoroughly outperforming random dropping and producing on-par (or better) results to early exit.
title Input Conditioned Layer Dropping in Speech Foundation Models
topic Sound
Computer Vision and Pattern Recognition
Audio and Speech Processing
url https://arxiv.org/abs/2507.07954