語音加密，不管是在軍事、商業或個人隱私上都是一個很重要的應用；其主要目地就是防止通訊時語音內容被不當的截取而造成資料外洩。一般而言，加密後之語音信號若具有殘留音訊(Residual intelligibility)，如無聲與有聲間的突然變化或出現與原聲相同的音節訊息等現象，將會造成語音加密安全性的降低。本論文提出兩類語音加密系統；一為類比語音加密，另一為數位語音加密。此兩類語音加密系統都是植基於正交分頻多工(OFDM)技術並利用擬隨機排序法，將語音資料隨機排序而達到語音加密的目的。

在類比語音加密系統中，我們發展出兩種不同架構；第一種稱為"使用樣本模式之OFDM語音加密系統"，此加密系統是利用擬隨機排序法直接將語音信號樣本隨機排序而達到語音加密的目的。由實驗結果得知，此種語音加密系統可有效達到語音通訊時之私密性，但其加密後之語音信號仍具有殘留音訊，因此會造成語音加密安全性的降低。另外，此加密系統可達到傳統FFT語音加密系統相同等級之安全性，但其系統架構卻比FFT語音加密系統簡單。第二種稱為"使用QAM模式之OFDM語音加密系統"，此加密系統是先將語音信號樣本轉換成二進制語音資料串，再利用64-QAM調變方式將此資料串映射至複數平面上，形成不同之複數符碼(Complex-valued symbol)，然後再利用擬隨機排序法將複數符碼隨機排序而達到語音加密的目的。由實驗結果得知，其加密後之語音信號呈現擬白色雜訊信號特徵，完全沒有殘留音訊，故可提供非常好之語音加密效果且安全性也非常高；而其解密還原後之語音信號品質與原始語音信號品質相差無幾。

在數位語音加密系統中，我們利用ITU-T G.729語音編碼器將語音信號壓縮編碼，其位元率為8 kbps，即每10 ms傳送80個二位元資料。由於編碼後代表語音參數之位元串，在傳送時若發生錯誤，其影響語音信號品質的程度不會完全相同。因此我們也發展出兩種不同的加密模式，一為全加密(Full encryption)，另一為選擇性加密(Selective encryption)。在全加密模式中，利用擬隨機排序法直接將80個二位元資料隨機排序而達到語音加密效果。在選擇性加密模式中，根據語音參數對語音信號品質影響程度的大小，適當地從80個二位元資料中選出36個較具影響性的二位元資料並給予加密處理，稱為高選擇性加密(High-selective encryption)；或是從80個二位元資料中選出24個較具影響性的二位元資料並給予加密處理，稱為低選擇性加密(Low-selective encryption)。全加密模式可應用在無功率限制的通訊環境中以提供較安全之語音加密效果；而選擇性加密模式則可應用在功率限制的通訊環境中，如行動通訊等。由實驗結果得知，高選擇性加密僅利用全加密模式中36 (45%)個較具影響性的二位元資料給予加密處理，其加密後之語音信號具有非常低的殘留音訊，在主觀聽覺上完全無法辨認其語音內涵，其加密效果近似於全加密模式。另外，由實驗數據顯示，全加密及高選擇性加密之安全性與利用AES演算法之語音加密技術具相同等級。在低選擇性加密模式中，由於僅利用全加密模式中24 (30%)個較具影響性的二位元資料給予加密處理，其加密後之語音信號具有較高的殘留音訊，在主觀聽覺上雖無法辨認其語音內涵，但其安全性也相對地降低。此外，本文所提數位語音加密系統包括全加密及選擇性加密模式，在相同測試條件下，當通道訊雜比低於14 dB時，其雜訊免疫力明顯優於利用AES演算法之語音加密技術。

最後，我們也利用兩種客觀評估法，評測加密後及解密還原後之語音信號訊號品質，以佐證本文所提OFDM類比及數位語音加密系統之優越性能。

Speech encryption techniques have been widely used in the corporate and military sectors for some time now. In general, a speech encryption system retains any considerable residual intelligibility such as talk spurts and the original intonation in the encrypted speech. This makes it easy for eavesdroppers to deduce the information contents from the encrypted speech. In this thesis, we proposed two types of speech encryption systems: the analog and digital encryption systems. Both speech encryption systems, based on an orthogonal frequency division multiplexing (OFDM) technique, can effectively encrypt a speech signal by permuting several frequency domain components within a speech frame.
In analog speech encryption system, two types of encryption schemes are proposed. One is called the sample mode because it treats speech signal at a sample level; the other is called the QAM mode because it treats speech signal at a bit level followed by a QAM mapping method. The proposed encryption system using the sample mode is based on an OFDM technique. It can effectively reduce the residual intelligibility from the encrypted speech by permuting several frequency domain components. From the simulation results, the proposed OFDM-based encryption system using the sample mode can provide the same level of security as does the FFT-based encryption system, but it needs only two FFT operations instead of the four required by the FFT-based one. The encryption system using the QAM mode is based on the combination of an OFDM technique and an appropriate QAM mapping method. It can completely remove the residual intelligibility from the encrypted speech by permuting several frequency domain components. Simulation results indicate that the proposed system using the QAM mode can attain a high level of security while retaining the good quality in the recovered speech.
In digital speech encryption system, the bitstream produced by the ITU-T G.729 standard at 8 kb/s is partitioned into two classes; one, the most perceptually relevant is to be encrypted, the other, to be left unencrypted. In this way, two kinds of selective encryption modes are developed, a low-selective encryption mode (24 bits out of 80), aimed at preventing most kinds of eavesdropping and a high-selective encryption mode (36 bits out of 80), based on the encryption of the most perceptually important bits and meant to perform as well as full encryption of the bitstream. Simulation results indicate that the proposed speech encryption modes (both full encryption and high-selective encryption) can provide a high level of security as does the AES-based speech encryption scheme. Furthermore, the high-selective encryption mode covers about 45% of the bitstream and achieves content protection equivalent to that obtained by full encryption of the bitstream. In contrast, the security of the low-selective encryption mode is inferior to that of the AES-based encryption scheme due to its little about 30% of the bitstream to be encrypted. In addition, as the channel SNR is below the value of 14 dB, the noise immunity of the proposed OFDM-based digital encryption system, including the full encryption, the high- and low-selective encryption modes, is better than that of the AES-based encryption scheme.
Finally, we also take two objective measures to assess the quality of the recovered speech and the residual intelligibility of the encrypted speech. Experimental results show the effectiveness of the proposed OFDM-based speech encryption systems, including the analog and digital types.

[1] S. Sridharan, E. Dawson, and B. Goldburg, “Fast Fourier transform based speech encryption system,” IEE Proc. I, Vol. 138, No. 3, pp. 215-223, June 1991.
[2] R. V. Cox, D. E. Bock, K. B. Bauer, J. D. Johnston, and J. H. Snyder, “The analog voice privacy system,” IEEE ICASSP, Vol. 11, pp. 341-344, Apr. 1986.
[3] L. S. Lee, "A Speech security system not requiring synchronization," IEEE Comm. Magazine, Vol. 23, No. 7, pp. 42-55, July 1985.
[4] B. A. Forouzan, Introduction to Cryptography and Network Security, McGraw-Hill Companies, 2008.
[5] R. J. Sutton, Secure Communications: Applications and Management, England: John Wiley and Sons, 2002.
[6] B. Goldburg, S. Sridharan, and E. Dawson, "On the use of a frequency domain vector codebook for the cryptanalysis of analog speech scramblers," IEEE Int. Symposium on Circuits and Systems, Vol. 1, pp. 328-331, June 1991.
[7] H. J. Beker and F. C. Piper, Secure Speech Communications, London: Academic Press Inc Ltd., 1985.
[8] B. Goldburg, S. Sridharan, and E. Dawson, “Cryptanalysis of frequency domain analogue speech scramblers,” IEE Proc. I, Vol. 140, No. 4, pp. 235-239, Aug. 1993.
[9] ----, "The automated cryptanalysis of analog speech scramblers," Springer-Verlag, pp. 422-430, 1998.
[10] R. M. Milton, “A time and frequency-domain speech scrambler,” Proceedings on Communications and Signal Processing, COMSIG 1989, Southern African Conference, pp. 125-130, June 1989.
[11] S. Sridharan, E. Dawson, and B. Goldburg, "Speech encryption in the transform domain," IEE Electronics Letters, Vol. 26, No. 10, pp. 655-656, May 1990.
[12] ----, "Speech encryption using discrete orthogonal transforms," ICASSP-90, pp. 1647-1650, 1990.
[13] K. Sakurai, K. Koga, and T. Muratani, “A speech scrambler using the fast Fourier transform technique,” IEEE J. Select. Areas Comm., Vol. SAC-2, No 3, pp. 434-442, May 1984.
[14] L. S. Lee, G. C. Chou, and C. S. Chang, “A new frequency domain speech scrambling system which does not require frame synchronization,” IEEE Trans. Comm., Vol. COM-32, pp. 444-456, Apr. 1984.
[15] A. Matsunaga, K. Koga, and M. Ohkawa, “An analog speech scrambler using the FFT technique with high-level security,” IEEE ICC 88., pp. 1619-1625, June 1988.
[16] ----, “An analog speech scrambling system using the FFT technique with high-level security,” IEEE J. Select. Areas Comm., Vol. 7, No 4, pp. 540-547, May 1989.
[17] S. E. Borujeni, “Speech encryption based on fast Fourier transform permutation,” IEEE-ICECS Conference Proceeding, Vol. I, pp. 290-293, Dec. 2000.
[18] M. S. Ehsani and S. E. Borujeni, "Fast Fourier transform speech scrambler," IEEE Int. Symposium on Intelligent Systems, Vol. 1, pp. 248-251, 2002.
[19] E. Dawson, "Design of a discrete cosine transform based speech scrambler," IEE Electronics Letters, Vol. 27, No. 7, pp. 613-614, Mar. 1991.
[20] F. Ma, J. Cheng, and Y. Wang, “Wavelet transform-based analogue speech scrambling scheme,” IEE Electronics Letters, Vol. 32, No. 8, pp. 719-721, Apr. 1996.
[21] B. Goldburg, S. Sridharan, and E. Dawson, “Design and cryptanalysis of transform-based analog speech scramblers,” IEEE J. Select. Areas Comm., Vol. 11, No. 5, pp. 735-744, June 1993.
[22] V. Senk, V. D. Delic, and V. S. Milosevic, “A new speech scrambling concept based on Hadamard matrices,” IEEE Signal Processing Letters, Vol. 4, No. 6, pp. 161-163, June 1997.
[23] V. Milosevic, V. Delic, and V. Senk, “Hadamard transform application in speech scrambling,” IEEE 13th Int. Conf. on DSP, Vol. 1, pp. 361-364, July 1997.
[24] Y. Wu and B. P. Ng, "Speech scrambling with Hadamard transform in frequency domain," IEEE ICSP'02 Proceedings, Vol. 2, pp. 1560-1563, Aug. 2002.
[25] G. Manjunath, and G. V. Anand, “Speech encryption using circulant transformations,” IEEE Int. Conf. on Multimedia and Expo., Vol. 1, pp. 553-556, Aug. 2002.
[26] Q. H. Lin, F. L. Yin, T. M. Mei, and H. L. Liang, "A speech encryption algorithm based on blind source separation," IEEE ICCCAS, Vol. 2, pp. 1013-1017, June 2004.
[27] ----, “A Blind source separation based method for speech encryption”, IEEE Trans. On Circuits and Systems-I, vol. 53, No. 6, pp. 1320-1328, June 2006.
[28] National Institute of Standards and Technology (NIST), " FIPS-197: Advanced Encryption Standard", Nov. 2001.
[29] W. Zou and Y. Wu, "COFDM: An overview," IEEE Trans. on Broadcasting, Vol. 41, No. 1, pp. 1-8, Mar. 1995.
[30] R.V. Nee and R. Prasad, OFDM for Wireless Multimedia Communications, Artech House, Boston, 2000.
[31] A. Pandharipande, "Principle of OFDM," IEEE Potentials, Vol. 21, pp. 16-19, Apr. 2002.
[32] J. Heiskala and J. Terry, OFDM Wireless LANs: A Theoretical and Practical Guide, Sams Publishing, 2002.
[33] L. Hanzo, M. Münster, B. J. Choi, and T. Keller, OFDM and MC-CDMA for Broadband Multi-User Communications, WLANs and Broadcasting, John Wiley & Sons, Ltd., England, 2004.
[34] A. R. S. Bahai, B. R. saltzberg, and M. Ergen, Multi-Carrier Digital Communications Theory and Applications of OFDM 2nd, Springer, 2004.
[35] R. J. Schilling and S. L. Harries, Fundamental of Digital Signal Processing Using MATLAB, Nelson, Thomson Ltd., 2005.
[36] ITU-T, “Recommendation G.729: Coding of speech at 8 kbit/s using conjugate structure algebraic-code-excited linear-prediction (CS-ACELP),” Mar. 1996.
[37] R. Salami, C. Laflamme, J. P. Adoul, A. Kataoka, S. Hayashi, T. Moriya, C. Lamblin, D. Massaloux, S. Proust, P. Kroon, and Y. Shoham, "Design and description of CS-ACELP: a toll quality 8 kb/s speech coder," IEEE Trans. on Speech and Audio Processing, Vol. 6, No. 2, pp. 116-130, Mar. 1998.
[38] D. C. Tseng and J. H. Chiu, "A new speech scrambler based on OFDM technique," IJEE Trans. Vol. 13, No. 4, pp. 341-348, Nov. 2006.
[39] S. E. Borujeni, "Cryptography by pseudo random number generator," IEEE Int. Symposium on Intelligent systems, pp. 244-247, Sep. 2002.
[40] D. C. Tseng and J.H. Chiu, "An OFDM speech scrambler without residual intelligibility," IEEE TENCON 2007, pp. 122(4), Nov. 2007.
[41] ----, "An OFDM-based speech encryption system without residual intelligibility," to be appeared in IEICE Transactions.
[42] J. J. van de Beek, M. Sandell, and P. O. Borjesson, "ML estimation of time and frequency offset in OFDM systems," IEEE Trans. on Signal Processing, Vol. 45, No. 7, pp. 1800-1805, July 1997.
[43] G. Yang, P. Ho, and V. Cuperman, "Error protection for a 4.8 kbps VQ based CELP coder," IEEE Conf. Vehicular Technology, pp. 726-731, 1990.
[44] K. Swaminathan, A. R. Hammons Jr., and M. Austin, “Selective error protection of ITU-T G.729 CODEC for digital cellular channels,” IEEE ICASSP, pp. 577-580, 1996.
[45] A. Servetti and J. C. De Martin, "Perception-based partial encryption of compressed speech," IEEE Trans. on Speech and Audio Processing, Vol. 10, No. 8, pp. 637-643 Nov. 2002.
[46] J. D. Gibson, A. Servetti, H. Dong, A. Gersho, T. Lookabaugh, and J. C. D. Martin, "Selective encryption and scalable speech coding for voice communications over multi-hop wireless links," IEEE MILCOM-2004, pp. 792-798, 2004.
[47] T. Lookabaugh and D. C. Sicker, "Selective encryption for consumer applications," IEEE CCNC, pp. 516-521, 2004.