8/2/2019 05705790

1/5

Abstract - Steganography is the art of hiding message inorder to have a secure data communication. This paper

addresses a technique for wave steganography. In this paper

we proposed the idea to replace bits according to the

distortion afforded with lossy or lossless and recovery

methods. Carrier file bits are replaced by message file.

Message embeded in this method is in form of wave.

Hidden message can be recovered by applying reverse

algorithm. Message file is hidden into the carrier file by

using LSB technique, as modifications will usually not createaudible changes to the sounds.

Results show that the proposed methods are novel for

wave steganography for 16 bit stereo wave files.

.

Keywords - Information hiding, Steganography, LSB.

I. INTRODUCTION

Steganography, coming from the Greek words

stegos meaning roof or covered and graphia which means

writing, is the art and science of hiding the fact that

communication is taking place. Using steganography, youcan embed a secret message inside a piece of unsuspicious

information and send it without anyone knowing of theexistence of the secret message. The prevalence of

multimedia data in our electronic world exposes a new

avenue for communication using digital steganography.

Steganography, where the occurrence of communicationis concealed, differs from cryptography, in which

communication is evident but the content of that

communication is camouflaged. To be useful, a

steganographic system must provide a method to embed

data imperceptibly, allow the data to be readily extracted,

promote a high information rate or payload, and

incorporate a certain amount of resistance to removal [1],[2].

The two primary criteria for successful embedding of a

covert message are that the stego signal resulting from

embedding is perceptually indistinguishable from the host

audio signal, and the embedded message is recovered

correctly at the receiver. Other requirements such as

robustness of embedding, data recovery without the

original carrier signal, etc. may depend upon the type of

applications [3].

In general, direct extension of the bit modification

technique to host audio signals is precluded by the higher

sensitivity and dynamic range of the human auditory

system (HAS) [4], compared with the visual system. With

a large power and dynamic range, the human ear can

detect a change in an audio file as low as one part in 10

million.

If the embedded message is another audio signal, one

may perceive the intended message even if some errors

are encountered during the recovery. This is particularly

important in covert communication as the stego signalmay he received with intentional and unintentional

changes due to noise in the channel or attacks. The

primary goal here is to convey the concealed messagealbeit with reduced speech quality.Additionally, some degradation in the perceptual

quality of the stego signal from that of the original host

signal may be acceptable. If the host used for carrying the

covert message is not a common or familiar audio signal,

the degraded quality of the stego signal may not be

noticeable by attackers; hence, the presence of

information hidden in the stego may stay imperceptible

and, consequently, impervious. Based on this premise, amethod of data embedding by manipulation of the host

samples in accordance with the covert information is

described in this paper.

For CD quality sound [13], it is necessary to encode16 bits per sample at a rate of 44.1 kHz. Since the

majority of the sound signal is stored in the M most

significant bits of the sound code, changing the lowest N

significant bits leaves the sound signal somewhat changed

in quality and definition, but still recognizable, and often

indistinguishable from the original.

Data hiding in the least significant bits (LSBs) [18] of

audio samples in the time domain is one of the simplest

algorithms with very high data rate of additional

information [19,20]. LSB coding [5,6] is one of the

earliest techniques studied in the information hiding.This paper proposes two techniques for wave

steganography. First is lossless hiding and recovery,where message file doesnt change. Second is lossy,

where message file somewhat changes but offer high

capacity embedding (up to 100% of the carrier file) in

carrier file.

II. RELATED WORKS

In this section review of the works which has been doneon audio steganography is presented. Time domain is oneof the common domains used for steganography [7]. Time

domain steganography is usually simple and fast [8]. Forexample wavelet domain [9] and Fourier domain [10] are

used to hide information in audio signals.

Wave Steganography

Anant Umbarkar1, Abhijit Joshi2, Ajay Jadhav31Department of Information Technology, W.C.E. Sangli, Maharashtra, India

2 Tata Consultancy Services, Ahmadabad, Gujarat, India3Infosys, Kolkata, India

e-mails: 1anantumbarkar@rediffmail.com, 2abhijitpjoshi@gmail.com, 3jadhav.ajay11@gmail.com

978-1-4244-5967-4/10/$26.00 2010 IEEE

8/2/2019 05705790

2/5

Different domains have special features which made

them suitable for different application. But when thehiding and unhiding procedures are done in another

domain such as wavelet or Fourier, some error are

introduced. The source of this error is usually rounding

error occurred during transform or when the signal is

saved as a file or transferred via a communicationchannel. So it must be considered to know the error rate

[11]. There are Error Correction Coding (ECC) methods

which can be used to achieve zero error rates [12], but it ismore desirable to have a method with low error rate.

III. SAMPLE DATA

The methods proposed supports 16 bit stereo wave files.

16 bit stereo wave file has two channels with each sample

of 16 bits. Each sample of wave signal is in the range of

-1 to +1 and is double values. Table I shows the sample

wave files used. Carrier.wav is a carrier wave file into

which a message file M1F1-int12-AFsp.wav is to behidden. And after hiding the new file called Stego.wav is

created. Size of carrier file is 113KB, size of M1F1-int12-

AFsp.wav is 91.9KB and 113 KB of Stego.wav file.

TABLE ISAMPLE WAVE FILES USED

Type Name Size

Carrier carrier.wav (Stereo) 113 KB

Message M1F1-int12-AFsp.wav (Mono) 91.9 KB

Stego Stego.wav 113 KB

IV. IMPLEMENTATION

Numerous steganographic techniques have been

presented, what is due to fact that multimedia objects

have a highly redundant representation. The most

common and simplest steganographic method is the least

significant bit (LSB) insertion method that embedsmessage in the least-significant bit of host audio.

Method replaces the least significant bits according to

the percentage of the data to be hidden. The number of

bits replaced in hiding is called the numbits. Message canbe hidden in the carrier file by using two methods.

A. Lossless method

This method replaces least significant bits of the

carrier file by bits of the message file for all the bits of the

message file as we dont want to loose any information in

lossless mode.It is used where importance given to the data and

dont want to loose any information associated with the

carrier file.

Steps carried out in lossless method

1. Number of bits to be replaced i.e. numbits isselected.

2. Read the carrier and message files.3. Filter the sample values by following method asshown in fig. 1.

filtered=int - mod(int, 2^numbits);

Where,

int = the integer value of the sample

calculated by shifting the bits.filtered = filtered sample value.

4. Convert the double valued message samples tointegers.

5. These integer values encrypted and thenconverted to binary values.

6. Select most significant numbits bits of themessage file and convert to decimal value.

7. Add this decimal value to the filtered value(filtered) by following formula:

tmp = filtered + dec;

where,dec = decimal value calculated in step 5.

tmp = temporary combined integer value.8. Convert value of tmp back to double and save.9. Repeat step 5, 6, 7 until all 16 bits of message

channel are selected then go to next sample.

At the receiver side to recover the embedded message file

from the carrier file reverse algorithm is used.

Fig. 1. Lossless channel 1 for numbits=8

Fig. 2. Lossless channel 2 for numbits=8

Fig. 1 and 2, shows the hiding technique for

channel 1 and channel 2 respectively for numbits=8. In

this, the method filtered the sample of carrier channe1and

places the first 8 bits of the message data in the filtered

place.

8/2/2019 05705790

3/5

B. Lossy method

This method replaces least numbits bits of the carrier file

by numbits bits of the message file for only numbits bits

up to 8 most significant bits of the message file Since the

majority of the sound signal is stored in the mostsignificant bits of the sound code.

Steps carried out in lossless method

1. Number of bits to be replaced i.e. numbits isselected.

2. Read the carrier and message files.3. Filter the sample values by following method asshown in fig. 1.

filtered=int - mod(int, 2^numbits);

Where,

int = the integer value of the sample

calculated by shifting the bits.filtered = filtered sample value.

4. Convert the double valued message samples tointegers.

5. These integer values encrypted and thenconverted to binary values.

6. Select most significant numbits bits of themessage file and convert to decimal value.

7. Add this decimal value to the filtered value(filtered) by following formula:

tmp = filtered + dec;where,

dec = decimal value calculated in step 5.

tmp = temporary combined integer value.8. Convert value of tmp back to double and save.9. Repeat step 5, 6, 7 until 8 MSB bits of message

channel is selected then go to next sample.

At the receiver side to recover the embedded message file

from the carrier file reverse algorithm is used.

Fig. 3. Lossy channel 1 for numbits=8

Fig. 4. Lossy channel 2 for numbits=8

Implementation of this proposed work is done in

Matlab 7.1.

Fig. 3 and 4, shows the hiding technique for channel 1

and channel 2 respectively for numbits=8. In this, themethod filtered the sample of carrier channe1and places

the first 8 bits of the message data in the filtered place andleast 8 significant bits are not considered.

C. Encryption of message

To add another tier of security we added encryption

algorithm. Simple XORing encryption is implemented before hiding message wave file in to the carrier wave

file.

D. Noise

As wave file is digital data we calculate noise in no.

of bits changed in stego wave file with respect to carrier

wave file. We found that the noise level in stego wave file

is negligible to carrier wave file.

V. RESULTS AND DISCUSSION

Fig. 5 carrier sample fig. 6 message sample fig. 7

stego sample and fig. 8 recovered sample show the plot ofamplitude versus time for lossless method in which

numbits = 8.

As shown in fig. 5 and 7, there is no change in

carrier and stego file samples, which are created after

embedding the message samples. As numbits = 8 andmethod is lossless 50% message hiding is achieved with

respect to carrier file size.

Fig. 9 shows the percentage of hiding in carrier file

with respect to no. of bits selected to be replaced (i.e.

numbit) for lossless method. Fig. 10 shows the percentage

of hiding in carrier file with respect to no. of bits selected

to be replaced (i.e. numbit) for lossy method.

8/2/2019 05705790

4/5

Fig. 5. Carrier samples

Fig. 6. Message samples

Fig. 7. Stego samples

Fig. 8. Recovered samples

Fig. 9. Lossless Trade-off (% hiding capacity versus numbits)

Maximum hiding & recovery:

TABLE IIRECOVERED FILES

Method Name of message wave SizeRecoveredfile size

Lossless M1F1-int16-AFsp.wav 92 KB 57 KB

Lossy M1F1-int16-AFsp.wav 92 KB 92 KB

Table II shows the sizes of message file recovered. In

lossless method the size of the recovered wave file is 57

KB and it is 92 KB in lossy method if 8 bits of the carrier

wave file are replacedLossless method has advantage that the data is

recovered as it was at the time of hiding. But the

limitation is hiding capacity is half of that of the lossymethod for same value of the numbits.

Fig. 10 Lossy Trade-off (% hiding capacity versus numbits)

Lossy method has advantage that the hiding capacity is

double of that of the lossless method for same value of the

numbits. But the limitation is that there is insignificant

loss of message data as least 8 significant bits are not

considered.

8/2/2019 05705790

5/5

VI. CONCLUSION

In this paper we present a simple and effective

wave steganography method on LSB technique. This

Method replaces the least significant bits according to the

percentage of the data to be hidden. This method has highcapacity for hiding data. Another advantage of the

technique is that the hidden information is extracted by

methods; lossless and lossy. In Lossless data hiding, the

proposed scheme employs data hiding and recovery of the

data without any damage or loss to the message data. For

lossy hiding, there is insignificant loss of message data to

achieve maximum hiding. Experimental results show the

proposed method has high transparency, full recovery and

demonstrates correctness of the recovered data.

Encryption is adding another tier of security.

Future scopes of work include advanced encryption to

add higher security for message. As carrier is

uncompressed this method has scope for hiding even

more data by compressing message before hiding into the

carrier. Wave as message can be compressed by using

vector quantization [16, 17] to achieve hiding more data

into carrier file.

REFERENCES

[1] W. Bender, D. Gruhl, N. Morimoto, and A. Lu,

Techniques for data hiding,IBM Syst. J., vol. 35, 1996.[2] I. J. Cox, J. Kilian, T. Leighton, and T. Shamoon , Secure

spread spectrum watermarking for images, audio and video,in Proc. IEEE Int. Conf. Image Processing, Lausanne,

Switzerland, Sept. 1996, vol. 111, pp. 243246.[3] I. Cox, M. Miller "Electronic watermarking: the first 50

years", Proc. 4th IEEE Workshop on Multimedia SignalProcessing, Cannes, France, October 2001, pp. 225-230.

[4] F. Hartung, M. Kutter "Multimedia WatermarkingTechniques",Proceedings of the IEEE, Vol. 87, No. 7, July

1999, pp. 1709-1107.

[5] C. Yeh, C. Kuo "Digital Watermarking through Quasi m-Arrays", Proc. IEEE Workshop on Signal ProcessingSystems, Taipei, Taiwan, October 1999, pp. 456-461.

[6] T. Cedric, R. Adi, I. Mcloughlin "Data concealment in audiousing a nonlinear frequency distribution of PRBS coded data

and frequency-domain LSB insertion", Proc. IEEE Region10 International Conference on Electrical and Electronic

Technology, Kuala Lumpur, Malaysia, September 2000, pp.275-278.

[7] W. Bender, D. Gruhl, N. Morimoto, and A. Lu, "Techniques

for data hiding," IBM Systems Journal, vol. 35, issue 3-4,

September 1996, pp. 313-336.[8] N. Cvejic and T. Seppanen, "Channel capacity of high bit

rate data hiding algorithms in diverse transform domains, Proceeding of IEEE international Symposium on

Communications and Information Technology (ISCIT 2004),

October 2004, vol.1, pp. 84- 88.[9]N. Taraghi - Delgarm, Speech Watermarking, M.Sc. Thesis,

Computer Engineering Department, Sharif University ofTechnology, Tehran, IRAN, May 2006.

[10]L. Gang, A.N. Akansu, and M. Ramkumar, "MP3 resistantoblivious steganography," Proceedings of 2001 IEEE

International Conference on Acoustics, Speech, and Signal Processing (ICASSP'01), May 2001 vol.3, pp. 1365-1368.

[11] D. Kirovski and H.S. Malvar, "Spread-spectrum

watermarking of audio signals,"IEEE Transaction on SignalProcessing, vol. 51, issue 4, April 2003, pp. 1020- 1033.

[12] P. Sweene, Error Control Coding (An Introduction), Prentice-Hall International Ltd., Englewood Cliffs, NJ,

1991.[13] Tag Vaughan, Multimedia Making it work, 5th edition,

TMGH, 2004.[14] Johnson, N. F. and Jajodia, S, Exploring steganography:

Seeing the unseen, IEEE Computer Magazine, pp. 26-34,February 1998.

[15] S. K. Bandyopadhyay, Debnath Bhattacharyya, PoulumiDas, S. Mukherjee, D. Ganguly, A Tutorial Review on

Steganography,IC3 Noida, pp. 106-114, August 2008.

[16] M. Barlaud, P. A. Chou, N. M. Nasrabadi, D. Neuhoff, M.J.T. Smith, J. W. Woods, Introduction to the Special Issueon Vector Quantization, IEEE Trans. on ImageProcessing - Special issue on Vector Quantization, Vo1.5,No.2, pp. 197-199,February 1996.

[17] Y. Linde, A. Buzo, R. M. Gray, An Algorithm for VectorQuantizer Design, IEEE Trans. on Communications,

Vo1.28, [7] N. M. Nasrabadi and R. A. King, Imagecoding using vector quantization: A review, IEEE Trans.Commun., vol. 36, pp. 84-95, 1980.

[18] A. J. Umbarkar, A. P. Joshi, A. A. Jadhav, A. R. Buchade,

Wave Steganography Approach by Modified LSB ,Proc.2ndInternational Conference on Emerging Trends in

Engineering and Technology, ICETET-09 , Nagpur, India,June 2009, pp. 862-865, IEEE Computer Society

[19] K. Gopalan, "Audio steganography by cepstrum althoughthe algorithm is applicable to other resolution modification,"

In Proc. IEEE Int. Conf. Acoustics, Speech, levels of wavelet

and Signal Processing, Vol. 5, pp. 481-484, March 2005.[20] W. Bender, D. Gruhl, N. Morimoto and A. Lu, Techniques

for data hiding,IBM Systems Journal, Vol. 35, Nos. 3 & 4,pp. 313-336, 1996.