Tampering Detection On Digital Audio Using Gabor Filterbank

Nowadays, audio editing tools can easily utilized to alter any digital audio signal. The original recorded conversation can be modified by inserting fake statement in order to twisted the context. Most of such tampering are difficult to identify by relying only on human hearing. Hence, a robust tool is required to help detecting tampered audio if present. In forensic community, it is known that digital traces exists on each audio signal due to characteristics of the acquisition device. Detecting the acquisition device information can be helpful for forensic practitioner to evaluate consistency of the recording. In this study, Gabor filterbank features are investigate to classify the microphone models while take into consideration the issue of identical model. This features are analyzed and compared among several microphones in the experiments. The result indicated that Gabor filterbank feature has great potential to localize forgery that present on digital audio recording.