Search papers, labs, and topics across Lattice.
3
0
4
3
Winning systems in the AT-ADD challenge achieved over 90% accuracy in detecting all types of audio deepfakes, showcasing the potential of innovative detection strategies.
Audio-language models struggle significantly with hidden evaluation tasks, with accuracy plummeting by nearly 12 percentage points on average, highlighting the challenge of true audio comprehension.
Existing audio deepfake detectors are sitting ducks outside the lab: AT-ADD introduces a challenge to push research towards real-world robustness and generalization across all audio types.
