This paper presents a novel layering approach of steganography which implements deception theory by utilizing three types of multimedia, in order to hide the data in video file. In the proposed technique first, audio stream of video file is analyzed and two features are extracted, active zone and silence zone. These features define method numbers where each method number corresponds to one cipher technique which exists in the literature. After selecting the ciphering method, we apply it on the plaintext. In the second layer, first we convert the ciphered text into arbitrary characters and hide the data using chaotic map by generating logo image. In the third layer, we multiply that image with orthogonal matrix which is selected on the basis of silence zones. In the fourth layer, we hide that logo image using DWT with security key. The security key is generated locally by using logistic map, and its initial condition is set on the basis of selected method number. Lastly we hide resultant image into the audio signal using wavelet transform and regenerate the video file. The proposed system is implemented in MATLAB 7.0. Here, we analyzed the SNR values of the regenerated signal and other statistical results to prove that onion steganography works better than silence interval technique, LSB technique, echo hiding technique, magnitude technique and spectrum technique in many aspects.