Perceptual Digest Function for Verifying Integrity in Audio Forensics

In this work we propose a function that allows to calculate a summary code from the parameters of a voice signal. This function is based on ordering of spectral coefficients obtained by means of the application of the Fast Fourier Transform (FFT), using a locally generated reference function (Gaussian random noise). The proposed method is oriented to the verification of integrity in forensic voice signals. The proposed methodology has a perceptual approach, which implies that the resulting code is maintained, even when modifications are made, particularly those that do not affect the sensitive content of the signal, such as re-quantization processes