Skip to main content

Table 3 Bona fide vs. synthetic accuracy on dataset \({\mathcal {D}_{\text {eval}}}\) for each synthetic speech algorithm

From: Synthetic speech detection through short-term and long-term prediction traces

  Bicoherence    STLT STLT + Bicoherence   
  512 256 128   512 256 128
A07 0.541 0.505 0.501 0.865 0.813 0.864 0.905
A08 0.693 0.627 0.591 0.951 0.955 0.955 0.954
A09 0.543 0.508 0.508 0.835 0.882 0.865 0.835
A10 0.534 0.516 0.504 0.511 0.492 0.487 0.493
A11 0.617 0.685 0.762 0.629 0.489 0.481 0.474
A12 0.547 0.524 0.511 0.509 0.504 0.498 0.487
A13 0.768 0.779 0.767 0.948 0.955 0.955 0.945
A14 0.718 0.708 0.726 0.882 0.916 0.906 0.880
A15 0.567 0.514 0.507 0.466 0.479 0.473 0.465
A16 0.544 0.516 0.509 0.872 0.833 0.871 0.908
A17 0.510 0.532 0.578 0.656 0.649 0.660 0.653
A18 0.515 0.534 0.537 0.869 0.849 0.843 0.849
A19 0.611 0.586 0.575 0.882 0.863 0.885 0.906
All 0.592 0.578 0.578 0.739 0.741 0.737 0.735