Abstract: Detecting partial deepfake speech is essential due to its potential for subtle misinformation. However, existing methods depend on costly frame-level annotations during training, limiting ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results