Audio watermarking : reduced distortion LSB coding

Abstract

Audio watermarking is the technique of hiding data in audio signals. It is increasingly important for many applications, and is an active research area. Watermark is an invisible structure which is embedded into the host signals. Watermark must be imperceptible in its host to prevent unauthorized removal, easily extracted by the owner, and robust to incidental and intentional distortions. Digital audio watermarking provides a special challenge because the human auditory system (HAS) is extremely more sensitive than the human visual system (HVS). In this thesis, the proposed algorithm uses a blind digital audio watermarking scheme using time domain operations. The LSB coding is one of the earliest techniques studied in the data hiding. The proposed audio watermarking algorithm reduces the embedding distortion of the host audio. By using this proposed algorithm, watermark bits can be embedded into higher LSB layers. This provides the increasing of robustness against noise addition. Moreover, the perceptual quality of watermarked audio of this proposed LSB method is higher than that of the standard LSB method. The main advantages of this proposed LSB method include that the watermarked bits inserted layer cannot be easily detected by the unauthorized person, increasing robustness, a very high watermark channel bit rate and a low computational complexity of the algorithm. This project attempted to achieve the data hiding in higher layers, higher perceptual quality and to reduce distortion. The robustness of LSB audio steganography can be increased by this proposed LSB coding method.