Unique Software System for Verification and Identification of Audio Recording Equipment

Usually audio recording equipment (e.g. a smartphone) is identified by the phone number or the individual number of the mobile device. However, the SIM card can be disposed of making such identification impossible. The Fractal system uses another method and identifies audio recording equipment through the individual characteristics of the background noise equipment itself produces. As long as the original device is still available it is possible to verify if the message recorded was made using it or not.

In two phonograms, special characteristics of the wavelet analysis distinguish the characteristics of the audio recording equipment. As studies and long-term tests of the Silentium software suite show, among these characteristics can be identified individual characteristics for the background noise of the specific audio recording equipment.

After the characteristics are selected, an integral comparison criterion is formed - the distribution of characteristics by fractal scales for each of the two phonograms. Audio analysis is done for small time intervals, usually from 10 to 30 ms, the duration of which is determined adaptively, depending on a number of factors. The final conclusion is made based on the proximity to the integral criteria.

At the core of the system is the built-in Noise Characteristics Selection module for audio recording equipment. This module analyses ambient noise characteristics, unique for each digital audio recording device.

The Fractal system also includes several special built-in databases designed for verification and identification of audio recording equipment, depending on the duration of the phonograms.

Automatic device recognition relies on error graphs. All probability characteristics in the form of graphs of errors of the first and second kind are based on the custom test sets designed using selected experimental data for various groups of audio recording equipment and phonograms of different lengths.

The effectiveness of the Fractal device recognition system depends on the length of the source recordings. The feasible audio file duration starts from 30 ms and the longer the audio is, the smaller is the probability of errors.

The probability of errors is calculated based on graphs of errors of the first and second kind for large audio recordings data sets with sampling frequency >= 44,100 Hertz.

The use of a special technology for comparing the full spectra of the ambient noise of the equipment in two phonograms at small time intervals allows the Fractal software to provide highly precise results.

Depending on the closeness of the characteristics of the equipment noise in audio files the probability of an error in identifying audio recording equipment is 0.01% or less.