**4.1 Noise free channel simulation**

Simulation results of typical experiments with the Wavelet based scrambler, and descrambler for an Arabic word spoken by women's voice '"evenning" are shown in figures (4) to (8), and Tables (1) to (2), using different wavelets and different levels.

Case Study:

Using (Haar) Wavelet , (db3) wavelet, Sym2 and Sym4 wavelet each one will be considered with three different levels for the **Arabic word**'" **evening** ".

Figure (4)shows the waveform, spectrum, and spectrogram of a sample original clear speech signal that represents an Arabic word " evening ".

Speech Scrambling Based on Wavelet Transform 51

Fig. 6. Comparison between original speech and scrambled Speech using PSD Estimates

Fig. 7. Descrambled Speech Signal Using Haar Wavelet With Level 1; (a) Waveform. (b)

orignal speech signal.

Spectrum. (c) Spectrogram.

Fig. (7) shows the waveform, spectrum , and spectrogram of the resulted descrambled speech signal, while Fig. (8) shows the comparison of the descrambled speech signal and the

Fig. 4. Original Speech Signal; a) Waveform. (b) Spectrum. (c) Spectrogram. Using Wavelet Transform (Haar) With Level 1

Figure (5) shows the waveform, spectrum, and spectrogram of the scrambled speech signal, while the comparison of the scrambled speech signal, that resulted from applying a wavelet transform of type (Haar) with a specified level (level 1)is shown in Fig. (6) .

Fig. 5. Scrambled Speech Signal Using Haar Wavelet With Level 1; (a) Waveform. (b)Spectrum. (c) Spectrogram**;** 

Fig. 4. Original Speech Signal; a) Waveform. (b) Spectrum. (c) Spectrogram.

transform of type (Haar) with a specified level (level 1)is shown in Fig. (6) .

Fig. 5. Scrambled Speech Signal Using Haar Wavelet With Level 1; (a) Waveform.

Figure (5) shows the waveform, spectrum, and spectrogram of the scrambled speech signal, while the comparison of the scrambled speech signal, that resulted from applying a wavelet

Using Wavelet Transform (Haar) With Level 1

(b)Spectrum. (c) Spectrogram**;** 

Fig. 6. Comparison between original speech and scrambled Speech using PSD Estimates

Fig. (7) shows the waveform, spectrum , and spectrogram of the resulted descrambled speech signal, while Fig. (8) shows the comparison of the descrambled speech signal and the orignal speech signal.

Fig. 7. Descrambled Speech Signal Using Haar Wavelet With Level 1; (a) Waveform. (b) Spectrum. (c) Spectrogram.

Speech Scrambling Based on Wavelet Transform 53

The figures (9) to (11), in each one, figure (a) represents spectrogram comparison between original speech signal and scrambled speech signal, while figure (b) represents the comparison between the spectrogram of the descrambled and original speech signal. Both figures are tested under the same level of the chosen Wavelet Transform-Type: Sym2.

a) b)

a) b)

Fig. 10. (a) The Comparison between Original Speech and Scrambled Speech Using

(b) The Comparison between Original Speech and Descrambled Speech Using

Fig. 9. (a) The Comparison between Original Speech and Scrambled Speech Using

(b) The Comparison between Original Speech and Descrambled Speech Using

Case study with **SNR = 15 dB**.

**sym2 / Level 1** 

**sym2 / Level 1** 

**sym2 / Level 2** 

**sym2 / Level 2**

Fig. 8. The Comparison Between Original Speech and Descrambled Speech.

Table (1) shows distance measure (SEGSNRs) for the scrambled speech, while Table (2) shows the (SEGSNRd) distance measure for the descrambled speech for different Wavelets and different decomposition levels.


Table 1. SEGSNRs (dB) for the scrambled speech, for each wavelet with a specific level.


Table 2. SEGSNRd (dB) for the recovered speech, for each wavelet with a specific level.

#### **4.2 Noisy channel simulation**

An evaluation of the proposed speech scrambling system with different signal to noise ratios from (5 dB up to 25 dB) was tested.

#### Case study with **SNR = 15 dB**.

52 Advances in Wavelet Theory and Their Applications in Engineering, Physics and Technology

Fig. 8. The Comparison Between Original Speech and Descrambled Speech.

and different decomposition levels.

**4.2 Noisy channel simulation** 

ratios from (5 dB up to 25 dB) was tested.

Level of

Level of

Table (1) shows distance measure (SEGSNRs) for the scrambled speech, while Table (2) shows the (SEGSNRd) distance measure for the descrambled speech for different Wavelets

> Decomposition 1 2 3 Haar -4.8732 -4.0857 -4.0907 Db3 -4.7673 -3.7751 -3.8147 Sym2 -4.8064 -3.7710 -3.6743 Sym4 -4.6620 -3.7125 -3.9071

Table 1. SEGSNRs (dB) for the scrambled speech, for each wavelet with a specific level.

Decomposition 1 2 3 Haar 310.26 305.67 303.12 Db3 15.85 17.46 9.59 Sym2 112.96 18.19 13.91 Sym4 12.88 10.07 13.03 Table 2. SEGSNRd (dB) for the recovered speech, for each wavelet with a specific level.

An evaluation of the proposed speech scrambling system with different signal to noise

The figures (9) to (11), in each one, figure (a) represents spectrogram comparison between original speech signal and scrambled speech signal, while figure (b) represents the comparison between the spectrogram of the descrambled and original speech signal. Both figures are tested under the same level of the chosen Wavelet Transform-Type: Sym2.

(b) The Comparison between Original Speech and Descrambled Speech Using **sym2 / Level 1** 

(b) The Comparison between Original Speech and Descrambled Speech Using **sym2 / Level 2**

Speech Scrambling Based on Wavelet Transform 55

Decomposition 1 2 3 Haar -5.007 -4.267 -4.249 Db3 -4.863 -3.950 -4.032 Sym2 -4.898 -3.928 -3.625 Sym4 -4.796 -3.932 -4.112 Table 5. SEGSNRs (dB) for the scrambled speech, for each wavelet with a specific level, with

Decomposition 1 2 3 Haar 13.025 12.852 12.993 Db3 10.073 9.218 7.133 Sym2 12.349 11.001 9.438 Sym4 9.472 8.002 7.707 Table 6. SEGSNRd (dB) for the recovered speech, for each wavelet with a specific level, with

Decomposition 1 2 3 Haar -4.892 -4.111 -4.110 Db3 -4.771 -3.793 -3.846 Sym2 -4.805 -3.784 -3.682 Sym4 -4.672 -3.741 -3.932 Table 7. SEGSNRs (dB) for the scrambled speech, for each wavelet with a specific level, with

Decomposition 1 2 3 Haar 23.088 22.852 22.99 Db3 13.933 13.693 9.184 Sym2 20.304 16.255 12.956 Sym4 12.273 9.824 10.641 Table 8. SEGSNRd (dB) for the recovered speech, for each wavelet with a specific level, with

Level of

Level of

Level of

Level of

SNR = 15 dB

SNR =1 5 dB.

SNR = 25 dB

SNR =25 dB.

Fig. 11. (a) The Comparison between Original Speech and Scrambled Speech Using **sym2 / Level 3** 

(b) The Comparison between Original Speech and Descrambled Speech Using **sym2 / Level 3** 

The results from such tests are shown in Tables (3) to Table (8). Table (3) shows the SEGSNRs distance measure for the scrambled speech, and Table (4) shows the SEGSNRd distance measure for the descrambled speech, with SNR = 5 dB. Each two tables corresponding to a specific SNR.


Table 3. SEGSNRs (dB) for the scrambled speech, for each wavelet with a specific level, with SNR = 5 dB.


Table 4. SEGSNRd (dB) for the recovered speech, for each wavelet with a specific level, with SNR = 5 dB.

a) b)

The results from such tests are shown in Tables (3) to Table (8). Table (3) shows the SEGSNRs distance measure for the scrambled speech, and Table (4) shows the SEGSNRd distance measure for the descrambled speech, with SNR = 5 dB. Each two tables

Decomposition 1 2 3 Haar -5.867 -5.383 -5.292 Db3 -5.711 -5.146 -5.293 Sym2 -5.781 -5.094 -5.017 Sym4 -5.723 -5.220 -5.360 Table 3. SEGSNRs (dB) for the scrambled speech, for each wavelet with a specific level, with

Decomposition 1 2 3 Haar 3.195 2.852 2.993 Db3 2.530 2.015 1.407 Sym2 2.911 2.481 2.151 Sym4 2.262 1.995 1.204 Table 4. SEGSNRd (dB) for the recovered speech, for each wavelet with a specific level, with

Fig. 11. (a) The Comparison between Original Speech and Scrambled Speech Using

(b) The Comparison between Original Speech and Descrambled Speech Using

**sym2 / Level 3** 

**sym2 / Level 3** 

SNR = 5 dB.

SNR = 5 dB.

corresponding to a specific SNR.

Level of

Level of


Table 5. SEGSNRs (dB) for the scrambled speech, for each wavelet with a specific level, with SNR = 15 dB


Table 6. SEGSNRd (dB) for the recovered speech, for each wavelet with a specific level, with SNR =1 5 dB.


Table 7. SEGSNRs (dB) for the scrambled speech, for each wavelet with a specific level, with SNR = 25 dB


Table 8. SEGSNRd (dB) for the recovered speech, for each wavelet with a specific level, with SNR =25 dB.

Speech Scrambling Based on Wavelet Transform 57

Goldburg, B.; Dawson, E. & Sridharan, S. (1993). Design And Cryptanalysis Of Transform-

Goldburg, B; Sridharan, S. & Dawson, Ed. (1993). Design And Cryptanalysis Of Transform-

Graps, A. (2011). An Introduction to Wavelet. IEEE Computational Science and Engineering,

Gray, R; Buzo, A. & Matsuyama, Y. (1980). Distortion Measures for Speech Processing. IEEE

Gray, A. & Markel, J. (1976). Distance Measures for Speech Processing. IEEE Trans.

Guo, Da & Lin Q (2010). Fast Decryption Utilizing Correlation Calculation for

Lee, L. (1985). A Speech Security System Not Requiring Synchronization. IEEE

Mascarin, A. (2000). Wavelet Toolbox-Featured Product. The Math Works Inc. , Natick, MA,

Mermoul, A & Belouchrani, A. (2010). A Subspace – Based Method for Speech Encryption.

Mosa, E.; Messiha, N. & Abd El-Samie, F. (2010). Encryption of Speech Signal with multiple secret keys in time and transform domain. Int. J. Speech Technol., 13. Pichler, F. (1983). Analog Scrambling By The General Fast Fourier Transform. Department of

Rao, N. & Homer, J. (2001). Speech Compression Using Wavelets. Electrical Engineering

Sadkhan, S. B, Khaged, N. H., & Al-Saadi, L. H. (2005). A Proposed Speech Scrambling

Sadkhan, S.; Falah, N. (2007). A Proposed Analog Speech Scrambler Based on

Sakurai, K; Koga, K. & Muratani, T. (1984). A Speech Scrambler Using The Fast Fourier

Sridharan, S.; Dawson, E. & Goldburg, B. (1990). Speech Encryption In The Transform

Sridharan, S.; Dawson, E. & Goldburg, B. (1991). Fast Fourier Transform Based Speech

Whei, W. & Iang, H. (2000). The Automated Cryptanalysis of DFT-Based Speech Scramblers. IEICE Transactions on Information and Systems, Vol. E83-D, No.12.

Domain. Electronics Letters, Queensland Univ. of Technol., Vol. 26.

Encryption System. IEE Proceedings-I, Vol. 138, No. 3.

System Based On Wavelet Transform And Permutation. IEEE Communication And

Parallel Structure of Wavelet Transform. M.Sc. Thesis, AlNahrain University,

Transform Technique. IEEE Journal on Selected Areas in Communications, Vol.

BSS-based Speech Encryption System. Sixth International Conference on Neural

Int. Conf. On Information Science, signal Processing and their applications (ISSPA

http://www.amara.com/IEEEwave/IEEEwavelet.html

Acoustics, Speech and Signal Proc., Vol. ASSP-24, No. 5.

Communications Magazine, Vol.23, no.7.

http://www.mathworks.com/products/wavelet.

Trans. Acoustics, Speech and Signal Proc., Vol. ASSP-28, No. 4.

Communications, Vol. 11.

Computation.

2010).

IRAQ.

SAC-2, No.3

System Science.

Thesis Project.

Signal Processing, Vol. 3.

Communications, Vol. 11, No.6.

Based Analog Speech Scramblers. IEEE Journal On Selected Areas in

Based Analog Speech Scramblers. IEEE Journal On Selected Areas in
