add audio branch embeddings

aiden200 / 2D3MF

Code and models for the paper "2D3MF: Deepfake Detection using Multi Modal Middle Fusion"

Other

31 stars 1 forks source link

add audio branch embeddings #1

Closed aromanusc closed 8 months ago

aromanusc commented 8 months ago

Adds Audio CNN feature embedding
- The audio embedding dimension for now is (batch_size, 8). The number 8 is an arbitrary number. Let's see how things fit in as soon as we start doing the transformer fusion.
Include utils.py to add audio utils
Cleanup code, remove comments

aromanusc commented 8 months ago

@aiden200 — I recommend rebasing your dev branch to this one to resolve potential merge conflicts. That way you will also pick up the audio development proposed in this PR.