Towards Robust Speech Models: Mitigating Backdoor Attacks via Audio Signal Enhancement and Fine-Pruning Techniques

Heyan Sun, Qi Zhong (), Minfeng Qi, Uno Fang, Guoyi Shi and Sanshuai Cui
Additional contact information
Heyan Sun: Faculty of Data Science, City University of Macau, Macau 999078, China
Qi Zhong: Faculty of Data Science, City University of Macau, Macau 999078, China
Minfeng Qi: Faculty of Data Science, City University of Macau, Macau 999078, China
Uno Fang: Curtin Institute for Data Science, Curtin University, Bentley, WA 6102, Australia
Guoyi Shi: Faculty of Data Science, City University of Macau, Macau 999078, China
Sanshuai Cui: Faculty of Data Science, City University of Macau, Macau 999078, China

Mathematics, 2025, vol. 13, issue 6, 1-16

Abstract: The widespread adoption of deep neural networks (DNNs) in speech recognition has introduced significant security vulnerabilities, particularly from backdoor attacks. These attacks allow adversaries to manipulate system behavior through hidden triggers while maintaining normal operation on clean inputs. To address this challenge, we propose a novel defense framework that combines speech enhancement with neural architecture optimization. Our approach consists of three key steps. First, we use a ComplexMTASS-based enhancement network to isolate and remove backdoor triggers by leveraging their unique spectral characteristics. Second, we apply an adaptive fine-pruning algorithm to selectively deactivate malicious neurons while preserving the model’s linguistic capabilities. Finally, we fine-tune the pruned model using clean data to restore and enhance recognition accuracy. Experiments on the AISHELL dataset demonstrate the effectiveness of our method against advanced steganographic attacks, such as PBSM and VSVC. The results show a significant reduction in attack success rate to below 1.5%, while maintaining 99.4% accuracy on clean inputs. This represents a notable improvement over existing defenses, particularly under varying trigger intensities and poisoning rates.

Keywords: neural network; backdoor defense; speech enhancement; fine pruning (search for similar items in EconPapers)
JEL-codes: C (search for similar items in EconPapers)
Date: 2025
References: View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.mdpi.com/2227-7390/13/6/984/pdf (application/pdf)
https://www.mdpi.com/2227-7390/13/6/984/ (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:gam:jmathe:v:13:y:2025:i:6:p:984-:d:1614132

Access Statistics for this article

Mathematics is currently edited by Ms. Emma He

More articles in Mathematics from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().