Covert communication model for speech signals based on an indirect and adaptive encryption technique

摘要

This study proposes an indirect speech encryption technique that applies the chameleon style. A chameleon adapts its skin colour according to its surrounding environment to protect itself from its enemies. The proposed model adapts the wavelet coefficients of a secret speech to any other speech signal coefficients to make them similar. Subsequently, the vector to the adaptation that involves the original positions of the secret speech samples is sent to the receiver instead of the encrypted message or cover speech. The proposed method does not send the encrypted contents of the secret speech (such as that in traditional encryption) nor extends the bandwidth of the transmitted messages (such as that in a steganographic system). The proposed technique has been tested on several speech signals, and the reconstructed speeches have produced sounds that are similar to the original speeches with a normalised correlation of over 95%. Moreover, the model is robust against plaintext and ciphertext attackers. The proposed technique can be used in applications that require high-level security, such as military and intelligence communications.