Search papers, labs, and topics across Lattice.
1
0
2
You don't need deep learning to achieve 93% accuracy in detecting deepfake audio – a well-tuned SVM with interpretable acoustic features gets you there.