Cell-phone identification from audio recordings using PSD of speech-free regions

Advancements in cell-phone related technologies have led to much broader usage of modern cell phones than mere talking devices used for making and receiving phone calls. The user-generated audio/video signals from cell phones can be very helpful in a number of forensic applications such as securing the information left behind at a crime scene. This paper presents a system for cell-phone identification from audio recordings. The proposed system uses estimate of power spectral density (PSD) of speech-free regions as the feature vector corresponding to each audio recording. Support Vector Machine (SVM) is then used for classifying these feature vectors. The performance of the proposed system is tested on a custom database of twenty-six cell phones of five different brands. The proposed system shows promising results with an average classification accuracy of 88% for classifying cell phones belonging to different manufacturers. The average classification accuracy is lesser when all the cell phones belong to the same manufacturer.