Using encoder artefacts for steganalysis of compressed audio streams

Abstract

The ISO/MPEG1 Audio Layer-3 (MP3) audio compression algorithm [30, 113] is probably one of the most recognised and far-reaching developments in the area of digital media processing. The MP3 format enables compression rates of about 1/10 of the size of uncompressed digital audio while degrading the audible quality only marginally. Together with faster computers the moderate complexity of the compression algorithm, due to which software implementations of MP3 encoders/decoders with acceptable performance even on low budget home computers soon became available, the format simplified the interchange of music. As a result, MP3 gained worldwide popularity among its users and at the same time it threatened the music industry’s conventional business models. The popularity of the format created demand for encoding tools and opened a market for a variety of programs for different needs. Today we count hundreds of MP3 encoder front ends based on several dozens of encoding engines, ranging from proof of concepts to targeted products either tuned for high speed, or optimised to costly and flexible tools for professionals.