Audio Watermarking Scheme in MDCT Domain

Abstract

In our proposed watermark embedding system, the original audio is segmented into overlapping frames. The psychoacoustic auditory model has been utilized to calculate the global threshold in modified discrete cosine transform domain. The perceptual insignificant locations have been used to insert the appropriately scaled watermark in the transform domain. Blind detection of watermark has been performed. Simulation results indicate that the proposed watermarking system is perceptually transparent and robust against various kind of attacks such as AWGN addition and MP3 compression.