Skip to main content

Table 5 Detection performance of feature matrix \({F_{m \times m}}\), \({P_{n \times n}}\)

From: Shallow and deep feature fusion for digital audio tampering detection

Feature matrix

Classical dataset

GAUDI-DI dataset

Accuracy (%)

F1-score (%)

Accuracy (%)

F1-score (%)

\(F_{m \times m}\)

93.07

93.73

79.55

78.33

\(P_{n \times n}\)

91.09

91.28

77.69

76.15

  1. Bold means the best performance of tampering detection in the same experimental setting