Master Thesis

FZJ-2024-04917

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png Audio Deepfake Detection using the Stationary Wavelet Transform

Pillath, N. (Corresponding author) ; Suarez, E. (Reviewer)FZJ*

2024

This record in other databases:

Please use a persistent id in citations: doi:10.34734/FZJ-2024-04917

Abstract: Recent advances in generative modelling have uncovered new avenues for creative expression, while also raising the potential of malicious misuse. Deepfakes pose significant risks to personal privacy, information integrity and cybersecurity. One possible countermeasure against deepfakes are robust detection models based on deep learning. In the context of audio, convolutional architectures have demonstrated good recognition rates. The raw audio data is usually transformed into a suitable format for convolutional neural networks. Current state-of-the art models employ Fourier or wavelet transforms. This thesis investigates the detection of audio fakes using deep convolutional networks that process stationary wavelet transform inputs. The results show functional detector models, but also reveal limitations of the stationary wavelet transform.

Note: Masterarbeit, Rheinische Friedrich-Wilhelms-Universität Bonn, 2024

---

Contributing Institute(s):

1. Jülich Supercomputing Center (JSC)

Research Program(s):

1. 5112 - Cross-Domain Algorithms, Tools, Methods Labs (ATMLs) and Research Groups (POF4-511) (POF4-511)
2. 5122 - Future Computing & Big Data Systems (POF4-512) (POF4-512)

Appears in the scientific report 2024

---

Database coverage:

---