Forensic Analysis of Digital Audio Recordings based on Acoustic Mains Hum

Abstract

ENF (Electrical Network Frequency), fluctuates instantaneously from its nominal value (50/60 Hz) depending on an increase or decrease in power consumption as against power production in the grid network. An ENF-sourced noise component is added into audio recordings where mains power sourced electromagnetic field or acoustic mains hum exists. With the use of this component, recording date and time of an audio file can be verified. In this work, existence and estimation of the acoustic mains hum sourced ENF noise in audio files is studied by analysing the acoustic noise emitted by several devices that are frequently used at home or in workplace. Detection of the file recording time truly is examined by computation of the similarity between the ENF signals estimated from the audio recordings and the reference ENF obtained with the help of a circuit that is connected to power grid network. The behaviour of dynamic microphone and electret microphone towards acoustic mains hum is investigated and the extent of acquiring the information about recording device type and settings from audio files is analysed. Besides, as part of this work, ENF estimation from videos on social media is also investigated.