Cross-Language Audio Spoofing Detection Using GenSpoof-KAN: Tackling English and Arabic Deepfakes

View at publisher Summarise with VIRAT
Authors: Vijay Ganesh M, K. Kavitha, K. Gangasiva, Gullipalli Jahnavi, K. Hanusha, N. Harsha Vardhan
Cited by: 0
References: 20

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Audio spoofing attacks, specifically deepfakes, are massively used these days to compromise the security of automatic speaker verification (ASV) based systems, leading to data breaches and financial scams. Existing audio spoofing countermeasures are not well-generalized and experience issues when detecting unknown spoofing attacks, including deepfake audio and adversarial attacks. Moreover, Arabic audio spoofing detection has been largely neglected in the spoofing detection domain, primarily due to the scarcity of Arabic audio spoofing datasets. This paper proposes a novel generalized approach, GenSpoof-KAN, capable of reliable detection of audio spoofing attacks in English and Arabic languages. The combination of deep and multi-view features, when used with the KAN classifier, improves framework generalizability and enhances the detection performance for unknown spoofed audio samples. The obtained results demonstrate the effectiveness and better generalization aptitude of our approach while detecting unknown spoofing attacks, including codec compressions, channel variations, and encoding artifacts.

Fields of Study & Subjects

Speech Recognition and Synthesis Speech and Audio Processing Biometric Identification and Security

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
2104–2109
Publication date
21 January 2026
Language
EN