Linear Prediction Coefficients based Copy-Move Forgery Detection in Audio Signal

With the advancement digitalization, the issue of digital multimedia security has become one of the essential research areas. Digital multimedia security includes the analysis and classification of many digital data such as audio, image and video, etc. The most important problem with audio data is the difficulty of detecting audio forgery as auditory and visually. Considering the studies in the audio forgery field, the lack in the number of research in this field is a problem emphasized by many other researchers. In this study, we have developed a copy-move forgery detection system for audio signals. Linear Prediction Coefficients (LPC) are obtained to detect copy-move forgery and audio forgery is detected using Pearson Correlation Coefficient (PCC). The study basically consists of 3 stages. 1) By using the Yet Another Algorithm for Pitch Tracking (YAAPT) method, the silent and unsilent regions in the audio signals are detected and the audio signals have been divided into segments. 2) LPC coefficients of 5, 10 and 15 are obtained from these segments. 3) The similarity between the coefficients is calculated by PCC. In the experimental part, the effects of different coefficients on the success of the study are analyzed and the results is given in tables.