Network marks of montage in audio recordings

Abstract

Very often forensic domain processes required an approval of authenticity in audio recordings presented as admissible evidence. Standard techniques to search for editing in audio materials are rather long and wearisome. The paper proposes a network platform as an effective instrument for solving the above mentioned problem. A specialized software tool was developed to transform given audio data into set of nodes and links according to the algorithms of natural visibility graph and horizontal visibility. A comparative analysis of the derived network structures was performed with the use of popular Gephi software product. The results demonstrate the first advances of network paradigm for detection of audio montage, in addition the examples of trivial signals of those point on possible existence of a marker - the metric that responds to sound recordings tampering.