Search papers, labs, and topics across Lattice.
Michigan Technological University
1
0
2
Informative layers in speech models cluster in depth zones, and selecting them can boost audio deepfake detection performance by 28% over previous methods.
