search

ljj898 / CMDFD-Dataset-and-Deepfake-Detection

7 stars 0 forks source link

Missing audio modality in W2L and W2L_Self subset of the dataset. #3

Open ijaz3797 opened 2 months ago

ijaz3797 commented 2 months ago

I appreciate your work on deepfakes. I am using your dataset for my audiovisual DFD framework. I couldn't find audio modality in W2L, and W2L_Self. Am I missing something here like adopting audio from another subset?

ljj898 commented 1 month ago

The audio is the same for the same ID pairs across different forgery types. You can use audio from subset of another type with the same name, or you can use the original audio. The ID after the underscore '_' is the audio ID of the original audio