printer friendly version

Audio steganography

In the past decade, steganography, or the art of concealing information within seemingly innocuous digital platforms and carriers, has emerged into a hot area of research. This is due to its increasing use by hackers and miscreants for covertly hiding information.

Although steganography has gained notoriety as an approach that is useful only for illegitimate applications, it could be extremely useful in application areas such as intellectual property (IP) protection and copyright marking. Steganography allows media content publishers to hide a copyright mark in the content they produce. If copies of the produced content are made and distributed illegally, the copyright mark would remain on the copies as it is hidden from the user. Publishers could then use the copyright mark to identify owners of an illegal copy.

To hide information in audio files, substitution techniques are typically used by modifying a bit or few bits of information in an audio sample. However, copyright marks hidden in audio samples using substitution could be easily manipulated or destroyed if a miscreant comes to know that information is hidden this way.

In an attempt to bolster the robustness of content hidden in audio samples using substitution techniques, researchers from the College of Science and Technology at the University of Technology in Malaysia have come out with a novel algorithm that hides bits of information in multiple, vague least significant bit (LSB) layers of an audio file. In a nutshell, the algorithm is capable of ensuring that the hidden content would not be leaked when the audio content is subjected to intentional attacks or manipulations.

Content hidden in audio files typically gets distorted or leaked when it is subjected to two types of scenarios – when an intentional attack is launched to reveal hidden information in the files and when the audio file is subjected to noise additions or signal processing manipulation. To ensure robustness against the first scenario, the researchers have designed the algorithm in such a way that it embeds information in bits other than LSBs, since LSBs are most susceptible to intentional attacks aimed at extracting hidden information. Essentially, the researchers have ensured that it would be extremely difficult for attackers to discover which bits are the carriers of the hidden information, as opposed to conventional algorithms that hide information only in LSBs. To further increase the ambiguity, the algorithm is designed to modify indistinct audio samples as opposed to adopting a predefined procedure.

To protect audio files against the second scenario mentioned above, the researchers have ensured that the algorithm embeds bits of hidden information in deeper layers of the audio file and alters other bits to decrease the error. Typically, if a random selection of audio samples is chosen for carrying hidden information, low-power additive white Gaussian noise (AWGN) would be introduced in the sample. AWGN is easily detectable by the human auditory system and hence, the scope of modifying bits for hiding information would be limited. To overcome this problem, the researchers have designed the algorithm to embed message bits in deeper layers, albeit in an intelligent manner in order to decrease the intensity of errors.

Share this article:

Further reading: