MP3 Audio watermarking using calibrated side information features for tamper detection and localization

Audio contents are frequently stocked up and transmitted in compressed formats. Among the many existing audio compression schemes, MPEG-1 Audio Layer III (MP3) is very popular and still widely spread. Therefore, this paper represents a tamper detection method for MP3 files based on watermarking operating directly in the compression domain using Huffman data and recompression calibration. This method is based on a set of side information features embedded in the MP3 audio file as a watermark and used later to decide if the frame is doubly compressed. The detection of doubly compressed frames or singly compressed frames makes the selection of tampered frames possible. We made experiments on a set of MP3 audio clips to evaluate our method. Experimental results validate the efficiency of our approach compared to other methods from the literature.