A Speech Cryptosystem Based on Chaotic Modulation Technique Mahmoud F. Abd Elzaher *1 , Mohamed Shalaby **2 , Yasser Kamal **3 , Salwa El Ramly *4 * Department of Electronics and Electrical Communications, Faculty of Engineering, Ain Shams University, Cairo, Egypt 1 8273@eng.asu.edu.eg 4 Salwa_elramly@eng.asu.edu.eg ** Department of Computer Science, Arab Academy for Science, Technology & Maritime Transport, Cairo, Egypt 2 myousef73@hotmail.com 3 dr_yasser_omar@yahoo.com Abstract: In this paper, an encryption approach for Speech communication based on direct chaotic modulation (non-autonomous modulation) is presented, in which speech signal is injected into one variable of the master system (using Lorenz system) without changing the value of any control parameter. This approach is based on the change of chaotic signal by injecting Speech samples into one variable in chaotic system and hence generating a new chaotic signal. The Speech signal is then extracted from the chaotic signal on the receiver side. Furthermore, a high dimension chaotic system is used, which increases the security of the encrypted signal. Non-autonomous modulation technique is suitable for securing real-time applications. A comparative study of approach and Speech masking technique is also presented. Experimental results show that non-autonomous methods give better performance than their chaotic masking counterparts when they are analyzed against Signal-to-Noise-Ratio (-38.55 dB vs. -35.51 dB), Segmental signal-to-Noise-Ratio (-38.91 dB vs. -35.84 dB), Log-Likelihood Ratio (0.89 vs. 0.80), and Correlation Coefficient Analysis (0.0345 vs. 0.021). Non-autonomous techniques overcome the chaotic masking break and considered more secure. Keywords: Encryption; Speech encryption; Chaotic Modulation; Non-autonomous modulation; Lorenz system. 1 INTRODUCTION Speech communication is in close relation with daily life, such as education, commerce, politics, e- learning and news telecasting. With the advancement of modern telecommunication and multimedia technologies, Modern Speech communication systems demand a huge amount of information to be exchanged across Social Networks and the Internet every day so the need for encryption and security has increased. The conventional cryptographic techniques may be efficient for the text data; however, they are unsuitable to the bulk data capacity. One of the techniques that provide fast and highly secure encryption methods is chaos-based techniques. Continuous cryptographic systems have been developed which use the synchronization between the transmitter and receiver to retrieve data transmitted through an insecure medium. The first generation of these systems is masking. A Speech masking technique based on Lorenz System is presented in [1, 2] which uses Lorenz equation to generate Chaotic Signals, these signals are used as a base carrier signal on which the information signal is modulated at the transmitter side. The information signal is then recovered at the receiver side. The method of masking has been shown to be insecure as there are various cryptanalysis methods [3] that make it possible to estimate the sender dynamics and decoding of the message signal. The second generation is the parameter and non-autonomous modulation techniques. Non- autonomous techniques were developed to overcome the chaotic parameter modulation break, which includes the return map, and adaptive observer [4]. Non-autonomous modulation is considered to be more secure than parameter modulation. The main goal of this paper is proposing a Speech encryption system that provides users with a high degree of confidence and key sensitivity, and preserving a good quality of the reconstructed speech signal by chaotic systems. In section 2, Chaos-based cryptography systems are discussed. In Section 3, a speech masking technique based on Lorenz System is presented. In Section 4, the proposed Egyptian Journal of Language Engineering, Vol. 4, No. 1, 2017 1