Skip to main content

Table 6 Detection performance of deep feature

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

Deep feature

Classical dataset

GAUDI-DI dataset

Accuracy (%)

F1-score (%)

Accuracy (%)

F1-score (%)

Frequency deep features

94.06

94.32

84.46

81.12

Phase deep features

86.14

85.21

78.88

75.28

Deep features fusion

95.05

95.36

86.96

83.19

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