Research published in the International Journal of Intelligent Engineering Informatics, investigates how information can be hidden in a type of sound file, known as a wav file, a technique known as audio steganography.
Steganography has been used for thousands of years to hide information. A message written in so-called “invisible ink” is a primitive example of steganography. In the computer age, information has been hidden in text, image, and audio files without leaving any visible or noticeable changes. Steganography involves concealing the information as well as the existence of that information so that only the intended recipient can find it.
Audio steganography is usually considered a cost-effective means to encrypt data over a network, as it has low noise distortion and can generally be embedded without being detectable. R. Ramyadevi and V. Poornima of the Department of Computer Science at the SRM Institute of Science and Technology in Tamil Nadu, India, have estimated the maximum number of characters, numbers, letters, and other symbols, that might be added to an audio file without altering its structure. But more importantly they have looked at the limit on audible distortion and disturbance of the bit rate when the file is played as a normal sound file on a media player. If there is obvious distortion, then the fact that the file was being used for audio steganography might be more apparent to a third party. The wav, known formally as the waveform audio file format, was developed in 1991 by IBM and Microsoft and is widely used on personal computers and other devices.
The team’s study shows that accuracy can be improved at low embedding levels and deliver an optimal peak signal-to-noise ratio while obfuscating information if the first, second, and third least significant bits (LSBs) of the audio file are employed in the steganographic processing. The team compared 8-bit and 16-bit pulse-code modulation (PCM) audio and used mean square error (MSE), mean absolute error (MAE), signal-to-noise ratio (SNR), and cross-correlation analysis to identify hidden text data within a given audio stream. The team writes that a 16-bit stereo wav file can carry just over 30000 characters with spaces without the presence of that added information being detectable to a third part unaware of its presence in the file. An 8-bit mono wav file can carry more than 8000 characters with spaces.
If the first three LSBs were used accuracy was 98 percent and a false alarm rate of less than 5 percent was seen. Of course, the robustness of the hidden message within the audio file is affected by the length of the text message that is hidden, the team reports. More added information would be more obvious to a third party checking files for this kind of message obfuscation in a collection of sound files.
Ramyadevi, R. and Poornima, V. (2022) ‘Utilisation of audible steganography to organise and analyse the text within WAV files’, Int. J. Intelligent Engineering Informatics, Vol. 10, No. 5, pp.397–410.