ADVANCED LSB TECHNIQUE FOR AUDIO STENOGRAPHY

Dhinaharan Nagamalai et al. (Eds) : CoSIT, SIGL, AIAPP, CYBI, CRIS, SEC, DMA - 2017
pp. 79– 86, 2017. © CS & IT-CSCP 2017 DOI : 10.5121/csit.2017.70409
ADVANCED LSB TECHNIQUE FOR AUDIO
STENOGRAPHY
Mohammed Salem Atoum1
, Mohammad M Alnabhan2
and Ahmad
Habboush3
1
Faculty of Science, Information Technology and Nursing,
Irbid National University, Irbid, Jordan
2
Faculty of Information Technology, Mutah University
3
Faculty of Information Technology, Jerash University
ABSTRACT
This work contributes to the multimedia security fields by given that more protected
steganography technique which ensures message confidentiality and integrity. An Advanced
Least Significant Bit (ALSB) technique is presented in order to meet audio steganography
requirements, which are imperceptibility, capacity, and robustness. An extensive evaluation
study was conducted measuring the performance of proposed NLSB algorithm. A set of factors
were measured and used during evaluation, this includes; Peak Signal to Noise Ratio (PSNR)
and Bit Error Rate. MP3 Audio files from five different audio generators were used during
evaluation. Results indicated that ALSB outperforms standard Least Significant Bit (SLSB)
technique. Moreover, ALSB can be embedding an utmost of 750 kb into MP3 file size less than 2
MB with 30db average achieving enhanced capacity capability.
KEYWORDS
MP3, LSB, Cryptography, Steganography.
1. INTRODUCTION
There is a continuous challenge in securing digital transmission between network nodes against
any form of penetration and intrusion. Ensuring security of information requires considering three
main components confidentially, integrity, and availability. This can be conducted through
techniques, described as steganography and cryptography [1].
Cryptography is the process of encryption and decryption of a digital data. However,
cryptography techniques are considered week or consume high resources. Steganography is
mainly based on covering digital data in a safe digital carrier [2]. Steganography is utilized for
hiding secret messages in ancient times [3]. According to [4], steganography can be described as
a method of hiding secondary information (e.g. file or message) within primary information,
known as the carrier or host, with no effect on the size of information and without causing any
form of distortion. The information is embedded within a media expressed as a bit stream, stego
signal or sequence [5].

80 Computer Science & Information Technology (CS & IT)
Watermarking is another technique that is used to insert watermark into host cover to protect
information such as copyright for hosts [6]. Steganography and watermarking usually embed
information in host media in a transparent manner [7]. However, considering watermarking, the
process requires compromising intentional attacks and preventing any cause of information
destruction by insuring robustness and protecting signals quality [6]. Watermarking is the most
suitable technique in scenarios where hidden information knowledge can result in information
manipulations [7].
The strength of steganographic technique is based on saving the data in the carrier medium
against attacks or alteration. Audio files are considered very suitable media providing different
compression rates and allow performing steganography in MP3 format. Audio stenographic
methods based on SLSB have gained continuous concern. However, this technique has limitations
in security, capacity and imperceptibility. In addition, to date, embedding messages after audio
compression has not been widely considered. Accordingly, this work investigates standard audio
steganographic techniques and addresses its weaknesses and presents an Advanced Least
Significant Bit (ALSB) technique in order to improve robustness and security of the standard
LSB algorithm.
2. STANDARD LEAST SIGNIFICANT BIT
Standard Least significant bit (SLSB) algorithm is considered simplest steganographic method
[8]. In SLSB, secret message and cover are converted to stream of bits. One or more bit of secret
massage are used to replace cover LSB bits. Afterwards, bits stream are sent to the receiver which
uses the same algorithm to extract the embded message [9]. SLSB uses least significant bits of the
cover and utilizes sequential embedding. This result in clear suspicion secret message location
within the cover files [10]. Hence, it is easier to detect and extract the secret message by the
attacker [11]. In order to enhance the efficiency of SLSB algorithm security, a generator
described as pseudorandom number (PN) is used [12]. However, the use of PN has incurred time
limitations, because using PN requires more time to operate.
A few research works have been conducted in the area of MP3 audio steganography more
specially while considering embedding after compression [11]. The cause might be the weakness
of this technique in achieving a good well expansion of information data steganography, and in
some cases results in low quality sound. The MP3 file is compression file that means is not
flexible as well as the size is less compared to other audio file types [12]. The Embedding secret
message by using after compression methods is able to create audio corruption. Two methods
after compression are used to embed secret message: embedding in header frames and embedding
in audio data.
2.1 EMBEDDING IN HEADER FRAMES
Before describing methods using header frames to embed secret message, MP3 file structure is
explained. MP3 file is divided into header frames and audio data frames. Techniques used for
encoding MP3 files are: constant bit rate CBR, average bit rate ABR and variable bit rate VBR.
These methods are expected to use padding bytes. Several methods have utilized unused bits in
header frames and padding bytes before all frames and between frames, in order to replace bits
from secret message. However, weaknesses of these methods include; limited capacity and

Computer Science & Information Technology (CS & IT) 81
security. Using padding stuffing byte technique [13], the sender converts empty information in
padding byte in the cover with secret message. However, the capacity of embedded secret
message depends on the size of padding byte, which was added in the cover file using encoding
methods: CBR, VBR and ABR. At the receiver side, information search within stego file is
applied to find the location of padding byte in the cover and extract the secret message. Unused
bits in header frames such as private bit, copyright bit, original bit, and emphasis bit can be used
to embed secret message without affects the quality of sound [13]. This technique replaces bit
from secret message with a bit in the header. However, using this technique it is easily to extract
the secret message from the header and change it from attackers.
Using Before All Frames (BAF) technique [14], researchers develop new technique to embed
hole secret message before the first frames in the header. The secret message with encrypted text
is embedded in a cover file, will have a maximum size of 15 KB, however the size will reach 30
KB without using encryption. This technique is better capacity compared with padding and
unused bit, but also is less security without using encryption method before embedding the secret
message. In addition, Between Frames (BF) methods divide the secret message before embedding
it into small size cover file [14]. This method depends on the size of secret message, and on the
spaces between frames of the cover file. The maximum size of secret message can be embedded
is not fixed, because it can expanded the size of the cover file. The advantages of BF technique
are high capacity and imperceptibility, but the disadvantages are less security and robustness. It
can be concluded that all methods of header frame are facing limited robustness against attackers
[14].
2.2 EMBEDDING IN AUDIO DATA:
Several methods have addressed security problems in embedding the secret message in audio data
using header frames. [14] presents a new method that embeds one, two, three or four bits from
MP3 file by replacing one or two or three or four bits from the secret message, described in text
format. The first byte from the cover file is selected randomly. Using this method, random
position in the cover file is chosen to start embedding the secret message. This is sequentially
repeated to embed the secret message in the cover file. The drawback for using this method is
limited robustness as well as the random position it was used is not permanent with a fixed size.
3. PROPOSED ALGORITHM
To address limitations of embedding algorithms after compression, this work introduces a new
technique in LSB. The algorithm is described as Advanced Least Significant Bit (ALSB)
technique, which is developed to increase the security of secret message, and improves the
method of embedding the secret message in the host file. The main problem in LSB is its
weaknesses against intentional or unintentional attacks on the secret message. In ALSB, the
sender side uses random position selection from initial 100 byte from the host file. Moreover, the
value of LSB and MSB is used to select the bit location required to be embedded in the secret
message. If LSB and MSB have the same value, ALSB uses 4 bits from the secret message in
order to embed from location two to five of each byte. Otherwise, the technique uses just two bits
from secret message to embed it in location two and three. This methodology has increased the
security of LSB. The ALSB algorithm pseudo code is discussed below:

Algorithm: Advanced Least Significant Bit Algorithm
1: // H is the host file and SM is the secret message and H,SM are the inputs.
2: // H' is the host file (H+SM) and H' is output
3: // beginning to read host file H from initial bit and save it in H'.
4: start
5: For i = 1 to Size of (H) do
6: { H'i ← Hi
7: }
8: // Create random position from earliest 100 byte in the host file by
using random generation method Rp
9: // H' is the input
10: // Rp is the output
11: For i=1 to 100
12: do
13: // choosing the random byte
14: { Rp = position ( i )
15: }
16: // begin to create H' by using ALSB technique to insert message blocks MB
17: For i= Rp to size of (H')
18: do
19: { For j=1 to L (MB)
20: do
21: { Read the LSB and MSB value from the byte}
22: if the LSB+MSB = 00 or 11 then
23: { embed MB from 2nd to 5th position }
24: else if LSB+MSB= 10 or 01 then
25: { embed MB from 3rd to 4th }
26: Go to next byte
27: }
28: }
After the sender side implements ALSB technique, the stego object is constructed. To evaluate
stego object before send it via internet, the PSNR and BER methods are used to introduce the
results of noise in stego object. At the receiver side, inverse method is applied to predict the
secret message from the stego object. This prediction is based on the secret information received
from safe channel.
4. MEASUREMENT AND EVALUATION
This section describes main measurement metrics used to evaluate the proposed NLSB in terms
of reliability, imperceptibility and performance. These metrics include peak signal-to-noise ratio
(PSNR), and Bit Error Rates (BER). PSNR is the statistical value (ratio) that compared a signal's
maximum power with power of the signal's noise, it logarithmic scale and expressed in decibels
(db) [15]. However, PSNR is the peak error measure. The PSNR is error metrics used for quality
measurement. The PSNR value of the new algorithm is compared with PSNR of the SLSB
algorithm. Low when PSNR value is high, this describes better quality [16]. The PSNR equation
is shown below:

Where MAX is the maximum differentiation of the inputs (host file sample) compared with stego
object in order to validate if the stego object holds secret data or not.
The second metric used is Bit Error Rates (BER) which measures bit errors in a file, beside the
summation number of bits required for the spread during a time period. The BER is the statistical
measures in telecommunication transmission, the ratio result is percentage of a bit including
errors comparing to entire bits. This measure is expressed as ten to a negative power. If the results
is low data rate that means is very high in a transmission [15]. In addition, the BER is a measure
of bit error probability. BER is considered more accurate while considering increased number of
bit errors and long interval. BER is also used to measure the successful recovery of the hidden
information. This will have its high effect in real communication channel where errors exists
retrieving hidden information. BER is computed as described in the following equation:
Where L is the length, H is host file and H' is stego object.
Table 1 describes audio generators used during the experiment and explains specifications of each
audio file including duration in minutes and size in Mbps. These audio clips were used in the
evaluation study to measure the effectiveness of the proposed ALSB technique comparing to
Standard LSB (SLSB) and XLSB techniques.
Table 1. Audio file specifications
As shown in table 2, proposed ALSB achieved high PSNR values comparing to SLSB [17] and
eXtended LSB (XLSB) methods [18] for all audio files. XLSB was presented and evaluated in
[18]. While performing a T-Test between PSNR values of ALSB and SLSB the result in
(p=0.00088), which indicates a significant difference between these PSNR values with an
advantage to ALSB. Moreover, the BER result confirmed that the proposed ALSB over
performed SLSB and XLSB. ALSB algorithm achieved the lowest BER values comparing to
other algorithms. T-test between BER values of ALSB and SLSB results in (p = 0.0000735103),
which ensures a significant difference between BER values with an advantage to ALSB.
Accordingly, the proposed ALSB achieved high performance and outperformed SLSB
algorithms.
Name of Audio
generator
Time
(Minute)
Size under
320kbps (MB)
Pop 3:10 8.9
Rock 3:40 9.9
Blues 3:45 10.7
Hip-hop 4:30 12.4
Dance 5:30 14.2
Metal 7:00 14.8

Table 2 PSNR and BER results
Figures 1 and 2 illustrate the PSNR and BER results for XLSB, SLSB and ALSB techniques.
Figure1. PSNR Comparison Results Figure2.BER Comparison Results
5. CONCLUSIONS
This paper has investigated audio steganography, particularly with respect to MP3 files after
compression. In this concern, a new algorithm known as Advanced least significant bit algorithm
(ALSB) was presented aiming to meet audio steganography requirements including;
imperceptibility, capacity, and robustness. The proposed ALSB enhances steganography
efficiency, by not embedding the message in every byte of audio file. Alternatively, the location
of the first bit to be embedded is selected randomly and remaining bits are embedded considering
odd and even byte values in the audio file.
ALSB algorithm is considered an extension of standard least significant bit (SLSB). SLSB holds
sufficient information about cover format bits which are not manipulated. This increased errors or
distortions. In this work, ALSB was implemented and evaluated with comparison to SLSB.
Measurements and results indicated that in ALSB achieved improved capacity and increased
PSNR values (as imperceptibility representative) comparing to other methods such as SLSB and
XLSB. In addition, ALSB has shown an increased robustness against attacks by applying BER.
Accordingly, experiments show that ALSB method achieves increased average of capacity,
improved imperceptibility and advanced robustness.
Name of
Audio
generator
XLSB
PSNR
SLSB
PSNR
ALSB
PSNR
XLSB
BER
SLSB
BER
ALSB
BER
Pop 67.0086 61.1675 69.0515 0.04 0.05 0.0025
Rock 66.758 61.9193 68.0656 0.038 0.041 0.0024
Blues 67.9554 62.4768 68.5194 0.037 0.04 0.0024
Hip-hop 58.8168 62.8794 59.9845 0.035 0.038 0.0022
Dance 65.2817 62.7883 66.4976 0.034 0.037 0.002
Metal 66.8681 62.9386 67.8172 0.031 0.034 0.0019

REFERENCES
[1] Lentij J., “Steganographic Methods”, Department Of Control Engineering And Information
Technology, Budapest University. Periodica Poltechnica Ser. El. Eng. Vol.44, No. 3–4, P. 249–258
(2000), Url: Http://Www.Citesseer.Ist.Psu.Edu/514698.Html.
[2] Katzenbeisser S., Peticotas F., “Information Hiding Techniques For Steganography And Digital
Watermarking”, Artech House Inc.2000.
[3] Petitcolas F.A, Anderson R.J., Kuhn M.G., “Information Hiding – A Survey”, Ieee, Special Issue On
Protection Of Multimedia Content: 1062-1078, July, 1999.
[4] Cacciaguerra S., Ferretti S., “Data Hiding: Steganography And Copyright Marking”, Department Of
Computer Science, University Of Bologna, Italy, Url: Http://Www.Cs.Unibo.It/~Scacciag/Home-
File/Teach/Datahiding.Pdf.
[5] Nedeljko C. (2004). Algorithms For Audio Watermarking And Steganography. Acta Universitatis
Ouluensis. Series C., 2004..
[6] Andres G. (2002). Measuring And Evaluating Digital Watermarks In Audiofiles. Washington Dc.
2002.
[7] Andres G. (2002). Measuring and Evaluating Digital Watermarks in Audio Files. Washington DC.
2002.
[8] Nematollahi, Mohammad Ali, Chalee Vorakulpipat, and Hamurabi Gamboa Rosales. "Audio
Watermarking." Digital Watermarking. Springer Singapore, 2017. 17-38.….
[9] Shrivastav, Vijay. "A study of different steganographic methods." Journal of Digital Integrated
Circuits in Electrical Devices 2.1 (2017): 1-6.
[10] Arab, Farnaz, and Mazdak Zamani. "VW16E: A Robust Video Watermarking Technique Using
Simulated Blocks." Multimedia Forensics and Security. Springer International Publishing, 2017. 193-
221.
[11] Atoum, Mohammed Salem. "A Comparative Study of Combination with Different LSB Techniques
in MP3 Steganography." Information Science and Applications. Springer Berlin Heidelberg, 2015.
551-560..
[12] Atoum, M. S., Suleiman, M., Rababaa, A., Ibrahim, S., & Ahmed, A. (2011). A Steganography
Method Based on Hiding secrete data in MPEG / Audio Layer III. Journal of Computer Science,
11(5), 184-188 [12] Deng, K., Tian, Y., Yu, X., Niu, X., Yang, Y., & Technology, S. (2010).
Steganalysis of the MP3 Steganographic Algorithm Based on Huffman Coding. Test, (1), 79-82
[13] L. Maciak And M. Ponniah And R. Sharma, "Mp3 Steganography", 2008.
[14] Atoum, Mohammed Salem, Subariah Ibrahimn, Ghazali Sulong, Akram Zeki, and Adamu Abubakar.
"Exploring the challenges of MP3 Audio steganography." In Advanced Computer Science
Applications and Technologies (ACSAT), 2013 International Conference on, pp. 156-161. IEEE,
2013. Bhattacharyya, S., Kundu, A., Chakraborty, K., & Sanyal, G. (2011). Audio Steganography
Using Mod 4 Method. Computing, 3(8), 30-38.

[15] Bhattacharyya, S., Kundu, A., Chakraborty, K., & Sanyal, G. (2011). Audio Steganography Using
Mod 4 Method. Computing, 3(8), 30-38.
[16] El-Bendary, Mohsen AM. "FEC merged with double security approach based on encrypted image
steganography for different purpose in the presence of noise and different attacks." Multimedia Tools
and Applications (2017): 1-39.
[17] AbedulsalamAlarood, Alaa, et al. "HIDING AMessage IN MP3 USING LSB WITH 1, 2, 3 AND 4
BITS." International Journal of Computer Networks & Communications (IJCNC) Vol.8, No.3, May
2016.
[18] Atoum, M.S, " MP3 audio steganography techniqu using extended least significant bit", Thesis (Ph.D
(Sains Komputer)) - Universiti Teknologi Malaysia, 2014.

ADVANCED LSB TECHNIQUE FOR AUDIO STENOGRAPHY

More Related Content

What's hot (20)

Similar to ADVANCED LSB TECHNIQUE FOR AUDIO STENOGRAPHY (20)

Recently uploaded (20)

ADVANCED LSB TECHNIQUE FOR AUDIO STENOGRAPHY