A secret communication method based on watermarking technique using genetic algorithm

This paper presents a method of embedding a digital audio watermark into another host audio signal to achieve secret communication. It provides a secure way to avoid audio watermark is accessed by some illegal interceptors. This method first represents host audio in critical band representation by applying DWPT (discrete wavelet packet transformation) from spatial domain into frequency domain. The VQ (vector quantization) is adopted here to obtain a set of index and codebook for secret communication to compress the secret audio for hiding. Afterwards, the digital watermarking approach is proposed to embed the index into host audio. Many channels are available for data hiding. In order to obtain an approximate optimal performance, we adopt the genetic algorithm to select the best channel sets so that the audio is able to hide watermark optimally (quality and payload). Finally, we verify the robust of watermarking by using MP3 compression. The experimental results show that our method has high payload and the ability to against the loss of MP3 compression.