IMPROVEMENT METHODS FOR AUDIO STEGANOGRAPHY

• Main Authors: Chao-Kuo Huang, 黃照摑
• Other Authors: Ching-Kuen Lee
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/b423ss

碩士 === 大同大學 === 通訊工程研究所 === 106 === In modern life, various media such as the internet are often used to transmit and receive personal or business information. This also means that transmitting information may be extracted by illegal hackers. Thus, in order to improve the privacy and security in information transmitting, the development of steganography has received increasing attention. In a word, the technique of steganography is to hide secret information in an ordinary cover medium to produce a secret-embedded medium that is very similar to the original cover medium, referred to as the stego medium. Carrying the secret information, the stego medium is transmitted with the benefit of not easily attracting attention and preventing illegal extraction. The legal recipient can extract the secret information from the received stego medium by the known agreement. There are many possible types of secret information and cover media, such as text, data, sound, image, etc. The research in this thesis concerns the improvement methods of steganography for audio signals. Specifically, we selected audio signals for both the secret information and the cover media. This thesis begins with a review of existing audio steganography methods, including methods based on the least significant bit (LSB), parity coding, echo hiding, phase coding, and spread spectrum technique. Based on the LSB technology, the objective of our research is to develop audio steganography techniques that can increase the secret-information payload while reduce the perceptible influence of the received stego medium. Based on the experiments results, the research of this thesis showed that when the secret information is embedded, the perceptible influence will vary with the amplitudes of the original cover media. Specifically, with larger amplitudes of the cover media, the perceptible influence is less significant, which means that more secret information bits can be embedded in the parts of the cover medium with larger amplitudes. The research used two kinds of indexes for amplitude judgment, one is the amplitude of each sample of the cover medium, and the other is the average amplitude of the duration of 0.01 seconds of the cover medium. The experiments results showed that the use of average amplitude can increase the secret-information payload with no more perceptible influence of the stego medium. With the embedded secret information, each sample of the stego audio signal may have a difference (within a certain extent) from the original cover signal, and may therefore induce perceptible influence. The larger the difference, the perceptible influence is more significant. To reduce the possible difference between the cover signal and the stego signal, this research developed an improved method that can reduce the range of differences by half. By using comparative hearing test, this research developed and proposed a variable-LSB steganography method for audio signals, with a set of rules of determining the acceptable number of LSB bits that can be embedded in samples of the cover signal with different amplitudes. In summary, this research achieved the goal of increasing the amount of secret information payload and reducing the perceptible influence of the audio stego signal.