Forgery detection based on reverberation time estimation in multiple bands

In this paper we propose a system that detects an audio intrusion by analyzing the temporal evolution of an important acoustic parameter, i.e. the reverberation time (RT). Considering the forensic context of our application, we chose a blind method for estimating the reverberation time, and we propose a multiband approach for the RT's estimation. Furthermore, we introduce a succession of post-processing operations of the RT estimates in order to facilitate the detection of intrusions.