TSM resilient audio watermarking using IMFs

With widespread of the Internet as well as the numerous applications that involve audio signals and systems, the issue of ownership proof become imperative. In this paper, we present an audio watermark scheme that is based on Empirical Mode Decomposition and Hilbert Huang Transform. The audio signal is decomposed into several mono-component signals or Intrinsic Mode Functions to serve as the addressee for watermark. The watermark is a pseudo random number which is added to the highest and lowest Intrinsic Mode Functions of the signal to generate what we call the modified IMFs which is in turn added to the remaining IMFs to produce the watermarked signal. Our experimental results show that the proposed method is robust against signal processing attacks such as MP3, time scale modification, resizing, and others. The obtained results meet the recommended imperceptibility signal to noise ratio.