Sparse deepfake detection promotes better disentanglement
This addresses the need for interpretable and efficient deepfake detection systems in speech processing, though it is incremental as it builds on existing methods.
The paper tackled the problem of deepfake detection in speech synthesis by introducing sparse representations in the last layer of the AASIST architecture, resulting in an EER of 23.36% on the ASVSpoof5 test set with 95% sparsity and improved disentanglement metrics.
Due to the rapid progress of speech synthesis, deepfake detection has become a major concern in the speech processing community. Because it is a critical task, systems must not only be efficient and robust, but also provide interpretable explanations. Among the different approaches for explainability, we focus on the interpretation of latent representations. In such paper, we focus on the last layer of embeddings of AASIST, a deepfake detection architecture. We use a TopK activation inspired by SAEs on this layer to obtain sparse representations which are used in the decision process. We demonstrate that sparse deepfake detection can improve detection performance, with an EER of 23.36% on ASVSpoof5 test set, with 95% of sparsity. We then show that these representations provide better disentanglement, using completeness and modularity metrics based on mutual information. Notably, some attacks are directly encoded in the latent space.