Fig. 4From: Shallow and deep feature fusion for digital audio tampering detectionAccuracy of phase and frequency features, shallow features (\({F_1}\), \({F_f}\)), fitting coefficient (\({P_{\mathrm{coe}}}\), \({F_{\mathrm{coe}}}\)), input of the convolutional (\({P_{n \times n}}\), \({F_{m \times m}}\)), deep phase and frequency featuresBack to article page