Localization of Forgery on Audio Clips Using GLCM Features and Mel Spectograms

With the rapid development of powerful audio editing software makes the forgery of the digital audio easy. Researchers have proposed methods to cope with audio authentication in recent years. In this study, we investigated forgery traces on Mel spectrogram with visual techniques for the first time in the literature. For this purpose, the method divides the Mel spectrogram image into overlapping sub-blocks and extracts features via the Gray-level Co-Occurrence Matrix from these blocks for each RGB color channel. The feature vectors are lexicographically sorted to make the similar vectors closer. Similarity among blocks gives a clue about forgery. Experiment results show that the proposed method gives better results for the detection of audio copy-move forgery compared to other studies in the literature.