Fake Speech Wild: Detecting Deepfake Speech on Social Media Platform
This addresses the challenge of cross-domain deepfake audio detection for social media users, but it is incremental as it builds on existing methods with new data and augmentation.
The paper tackles the problem of detecting deepfake speech on social media by introducing the Fake Speech Wild dataset and evaluating countermeasures, achieving an average equal error rate of 3.54% across evaluation sets.
The rapid advancement of speech generation technology has led to the widespread proliferation of deepfake speech across social media platforms. While deepfake audio countermeasures (CMs) achieve promising results on public datasets, their performance degrades significantly in cross-domain scenarios. To advance CMs for real-world deepfake detection, we first propose the Fake Speech Wild (FSW) dataset, which includes 254 hours of real and deepfake audio from four different media platforms, focusing on social media. As CMs, we establish a benchmark using public datasets and advanced selfsupervised learning (SSL)-based CMs to evaluate current CMs in real-world scenarios. We also assess the effectiveness of data augmentation strategies in enhancing CM robustness for detecting deepfake speech on social media. Finally, by augmenting public datasets and incorporating the FSW training set, we significantly advanced real-world deepfake audio detection performance, achieving an average equal error rate (EER) of 3.54% across all evaluation sets.