PulseAugur
LIVE 13:04:59
research · [1 source] ·
0
research

Wavelet Scattering Transform enhances speech deepfake detection accuracy

Researchers have developed a new family of feature extractors called WST-X, utilizing the wavelet scattering transform (WST) to improve speech deepfake detection. This approach combines the interpretability of hand-crafted features with the higher-level information capture of SSL features. Experiments on benchmarks like Deepfake-Eval-2024 demonstrated that WST-X significantly outperforms existing methods by effectively identifying subtle acoustic anomalies. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Introduces a novel feature extraction method that improves speech deepfake detection accuracy and interpretability.

RANK_REASON Academic paper introducing a novel method for speech deepfake detection.

Read on arXiv cs.CL →

COVERAGE [1]

  1. arXiv cs.CL TIER_1 · Xi Xuan, Davide Carbone, Wenxin Zhang, Ruchi Pandey, Tomi H. Kinnunen ·

    WST-X Series: Wavelet Scattering Transform for Interpretable Speech Deepfake Detection

    arXiv:2602.02980v2 Announce Type: replace-cross Abstract: In this work, we focus on front-end design for speech deepfake detectors, the component that determines the discriminative acoustic cues provided to the classifier. Existing approaches are primarily categorized into two ty…