Image hiding in speech spectrum

Methods applied to ensure privacy of digital data became essential in many applications. In this paper, we propose a Fast Fourier transform based steganography technique that utilizes wideband speech as a cover to hide image. Selected locations from high-frequencies of the magnitude speech spectrum are exploited to embed the digital image. Naturalness and intelligibility of the resulted (stego) speech are maintained by employing a partial rather than a total replacement on the selected frequency components. Our simulation results show that this approach maintains the robustness of the original speech and achieves a higher hiding capacity.