Optimization of MFCC parameters for mobile phone recognition from audio recordings

Recent researches in the domain of electronics and telecommunication have made available various low price chips into a single smaller device. In the similar fashion, mobile phones are now equipped with efficient audio, video and image recorders. With the extended facilities and ease of use, it is frequently used to take photographs, videos and audio recordings. As the use of these devices increased, its use in criminal activities also increased. In such a scenario, the available recording may help in detection of its source device from which it has originated. Source device identification from audio recording is one of the major problems in the multimedia forensics. In this paper, we perform a study for optimization of MFCC (Mel Frequency Cepstral Coefficient) parameters in device identification from audio recording.