AUDIO STEGANOGRAPHY USING LEAST SIGNIFICANT BIT METHOD FOR CONFIDENTIAL DATA EMBEDDING

International Journal of Multimedia & Its Applications (IJMA) Vol.17, No. 3, June 2025
DOI:10.5121/ijma.2025.17303 31
AUDIO STEGANOGRAPHY USING LEAST
SIGNIFICANT BIT METHOD FOR CONFIDENTIAL
DATA EMBEDDING
Muhd Izzul Haziq Rahim, Nazirah Abd Hamid1
, Siti Dhalila Mohd Satar, Mohd
Fadzil Abdul Kadir, Ahmad Faisal Amri Abidin @ Bharun, Mohamad Afendee
Mohamed and Mumtazimah Mohamad
Faculty of Informatic and Computing, Tembila Campus, Besut, Terengganu
ABSTRACT
As the demand for secure communication increases, traditional encryption techniques, though effective,
often expose the presence of sensitive information, making them vulnerable to detection. Audio
steganography offers a more discreet alternative by embedding hidden data within audio signals,
preserving the secrecy of the communication. Among existing methods, the Least Significant Bit (LSB)
technique is widely used due to its simplicity and high imperceptibility. However, conventional LSB
methods face limitations in embedding capacity, robustness, and decoding reliability, particularly when
handling large datasets. This study proposes a modified LSB approach that addresses these issues by
introducing a compact binary encoding scheme and a fixed End-of-File (EOF) marker to improve
efficiency and accuracy. The method involves embedding custom binary representations of secret data into
8-bit audio samples, with extraction aided by the EOF marker. Results demonstrate that the waveform and
spectrogram comparisons confirmed the imperceptibility of the modifications, with a 100% similarity score
between original and stego-audio. While the technique is lightweight and effective, it is still susceptible to
audio compression and signal manipulation. Future enhancements may include adaptive EOF patterns and
integrated encryption to increase resistance to tampering. Overall, this research contributes a practical
and efficient advancement to audio steganography, offering improved security for embedded data in digital
audio communication.
KEYWORDS
Audio Steganography, Least Significant Bit (LSB), Data Embedding, wave and spectrogram
1. INTRODUCTION
The word 'steganography' is derived from the Greek words for 'covered' or 'concealed' ('steganos')
and 'to write' ('graphein'), which reflects its core purpose: to communicate in secret without
anyone noticing [1]. Steganography provides an additional layer of security by embedding
information into another medium, such audio, enabling secret communication [2][3]. without
drawing attention to yourself in the face of growing threats to data privacy and security Audio
steganography is particularly effective because the human ear is relatively insensitive to slight
changes in audio, especially in certain frequencies or phase components. Information can be
embedded in an audio signal without significantly changing the audio quality by changing small
aspects of the signal, such as the phase of sound waves or the least significant bits (LSB) of each
audio sample [4].
Despite the effectiveness of the LSB technique, it suffers from low robustness and limited
capacity when embedding large datasets. Existing research does not adequately address these

32
limitations or provide enhanced LSB variants that improve performance while maintaining audio
quality. This paper aims to bridge this gap by introducing a modified LSB approach with
enhancements in binary encoding and message delimitation. This paper is structured as follows:
Section 2 reviews related works in audio steganography. Section 3 introduces the Least
Significant Bit (LSB) method. Section 4 presents the proposed methodology. Section 5 discusses
enhancements to the standard LSB technique using custom binary encoding and an End-of-File
(EOF) marker. Section 6 evaluates the results through waveform and spectrogram analysis and
finally, Section 7 concludes the study and outlines potential directions for future research.
2. RELATED WORKS
Audio steganography has been the focus of extensive research, with numerous techniques
developed to conceal data within audio signals. This section reviews key methods—particularly
the Least Significant Bit (LSB) approach—and compares their strengths, limitations, and
applicability to secure data embedding.
2.1. Audio Steganography Techniques
Researchers have explored various audio steganography methods that hide secret messages
within audio files, carefully using the limitations of human hearing to keep these messages
undetectable. Various algorithms and approaches have been proposed to achieve this. One
common method is Least Significant Bit (LSB), where data is hidden in the least significant bits
of audio samples, making the modifications nearly imperceptible [5][6]. Another technique,
Phase Coding, embeds information by modifying the phase of the audio signal, ensuring minimal
distortion. Spread Spectrum is another approach that distributes the secret message across the
entire audio spectrum, making detection and removal more challenging. Additionally, Echo
Hiding involves embedding data by introducing inaudible echoes into the original audio signal,
further enhancing the stealth of the hidden message [7][8]. These techniques, among others, play
a crucial role in secure data communication through audio steganography.
2.2. Least Significant Bit
The Least Significant Bit (LSB) technique is widely used for hiding information in digital media,
such as audio, video, and images, without significantly altering the content [9][10]. This approach
embeds secret data into the least significant bits of the cover medium, which typically contribute
the least to the overall visual or auditory quality of the content. Over time, this technique has
been extensively investigated, and a substantial body of research has been dedicated to
understanding its characteristics, limitations, and potential applications [9][10].
Digital audio files are represented as a sequence of samples, each with a specific value. For
example:
• In an 8-bit audio file, each sample is represented by 8 bits (e.g., 10110110).
• In a 16-bit audio file, each sample is represented by 16 bits (e.g., 1011011010110110).
The least significant bit is the rightmost bit in this binary representation. Modifying this bit
changes the sample value slightly, but this change is usually inaudible to humans. In an 8-bit
sample (e.g., 10110110), the last bit can be altered without perceptible change. For example,
modifying samples to encode binary "1010" results in minor, inaudible changes. Although
effective for small data payloads, traditional LSB methods struggle with capacity and detection
resistance.

33
Example input:
Audio samples (8-bit):
11001100, 10010101, 11100001, 10101010
Secret message in binary:
1010
Embedding:
Replace LSB of each sample with each message bit:
11001100 → 11001101 (message bit: 1)
10010101 → 10010100 (message bit: 0)
11100001 → 11100001 (message bit: 1)
10101010 → 10101010 (message bit: 0)
Output:
Modified audio samples:
11001101, 10010100, 11100001, 10101010
2.3. Audio Steganography
Several audio steganography methods have been explored, including Phase Coding, Echo Hiding,
and Spread Spectrum. Table 1 summarizes key recent works:
Table 1. Comparative analysis of audio steganography methods.
Author(s) & Year Method Advantages Limitations
Abood et al. (2022)
[11]
Enhanced LSB + Bit
Cycling
Increased robustness;
includes encryption
features
Higher complexity;
reduced speed
Sayed & Wahbi (2024)
[12]
Phase Coding
High imperceptibility
and minimal audible
changes
Lower capacity; more
complex
implementation
Aslantas & Hanilci
(2022) [7]
Comparative Study of
Multiple Methods
Broad evaluation of
LSB, Echo, Spectrum
techniques
Does not propose a new
technique; only
comparative
3. METHODOLOGY
In audio steganography, the methodology involves carefully embedding confidential data into an
audio signal in a way that ensures the data is secure, robust, and remains undetectable to
unintended listeners. Usually, the procedure is picking a suitable audio file (cover audio),
choosing on an embedding method, encoding the secret data, and then retrieving it when required
[11]. Evaluation metrics like wave and spectrogram are used to assess the effectiveness of the
steganographic method. A well-defined methodology in audio steganography ensures that the
embedded data remains undetectable to unintended listeners while being reliably extracted by
authorized recipients. This study follows a structured process:
1. Select appropriate WAV audio files (16-bit PCM, 44.1kHz).
2. Convert the secret message into a custom binary format.
3. Embed binary into LSBs of each 8-bit sample.
4. Append an EOF marker (11111111) to signal message termination.
5. Evaluate the stego-audio using waveform and spectrogram analysis.

34
3.1. Framework
In this section, we introduce the framework used in the audio steganography process, focusing on
how the confidential data is embedded into the audio signal. We break down the process into two
main components: the sender’s role in embedding the secret data into the audio file, and the
receiver’s role in extracting that data accurately.
3.1.1. Sender Part
Figure 1 demonstrates how sender part works. The user need to insert the secret message and
sound carrier before both input goes to binary converter. After that, embedding process happen
where every binary in secret message will be embedded into a Least Significant Bit of every 8-bit
frame in sound carrier and produce Audio Steganography. This output will be sent through
communication medium.
Figure 1. Framework of a sender
3.1.2. Receiver Part
Figure 2 demonstrates how receiver part works after the Audio steganography file arrived at
intended receiver. Audio steganography file will be converted to binary before extraction
happens. At extraction process, the system will take Least significant bits from sound carrier and
combine it together after finish. This process will produce secret messages and the original sound
carrier.
Figure 2. Framework of a receiver

35
4. MODIFIED LEAST SIGNIFICANT BITS TECHNIQUES
In this section, we explore enhanced techniques for embedding confidential data using the Least
Significant Bit (LSB) method. While traditional LSB methods have proven effective, they have
limitations in terms of capacity and robustness. To address these challenges, we propose
modifications to the standard LSB technique. These modifications, including the use of custom
binary encoding and End-of-File (EOF) markers, aim to improve both the efficiency and security
of the data embedding process.
4.1. Custom Binary
The custom binary technique converts characters into compact binary sequences, with each type
of character—such as digits and letters—being represented by unique markers. Each character is
encoded with a prefix marker: digits (0), uppercase (10), lowercase (11). This reduces space and
identifies character type. This technique only takes ‘fix bit’ instead of whole 8 bit and assign
marker bit in front of character bit.
For example, Table 2, number (0-9) have a constant first four bit ‘0000’ and the rest four bit is fix
based on value. So, we take the fix bit and then we assign marker bit ‘0’ at front of character
binary to uniquely identify it was a number.
Marker Bit = ‘0’
Example : 9
Normal : 00001001
Propose technique : ‘0’ + 1001 = 01001
Table 2. Number binary.
For alphabetical, there is only 5 bit which the value is fix while order 3 bit from most significant
bit (MSB) is constant. For marker bit we assign ‘10’ for uppercase and ‘11’ for lowercase to
uniquely identify them. We can see the example for uppercase letter in Table 3.
Marker Bit = ‘10’
Example : C
Normal : 01000011
Propose technique : ‘10’ + 00011 = 1000011

36
Table 3. Uppercase binary.
4.2. End-of-File (EOF)
An EOF marker (11111111) signals the end of the embedded message, improving decoding
accuracy. Its purpose is to avoid the decoder incorrectly interpreting random bits in the file as
part of the message, it makes sure the decoder knows when the message finishes. The EOF
marker is added to the end of the binary sequence once the secret message has been converted
into its unique binary representation. The binary data is read during the decoding process until it
reaches the EOF marker, at which point it ends.
5. RESULTS AND DISCUSSION
The results of applying the proposed LSB technique were evaluated based on factors like sound
quality, efficiency, stealthiness, and capacity, with a focus on how well the hidden data remained
imperceptible. A comparison with the others LSB technique is also conducted to highlight
improvements and potential limitations. The waveform comparison between the original and
steganographic audio files as in Figure 3 shows that the LSB-based audio steganography
technique successfully hides the secret data without causing noticeable distortions. The
waveforms are almost identical, which means that the changes made to the audio were minimal
and didn’t affect the overall sound quality.
Figure 3. Wave comparison
Figure 4 shows the spectrogram for both original and steganography audio by showing that the
frequency content of both the original and the steganographic audio is nearly identical. The
spectrogram is a visual representation of how the frequency energy is distributed over time, and

37
the fact that there are no significant differences between the two spectrograms indicates that the
LSB embedding did not introduce any distortion or noise into the frequency spectrum.
Figure 4. Spectrogram comparison
One of the key advantages of the proposed modified LSB technique lies in its straightforward
design, which makes it well-suited for environments where system resources are limited. The use
of least significant bit embedding, paired with a compact binary encoding strategy, ensures that
the changes made to the audio file are subtle and generally imperceptible to the human ear.
Furthermore, incorporating an End-of-File (EOF) marker helps streamline the retrieval process
by clearly indicating where the hidden message ends, thereby reducing potential extraction errors.
However, there are limitations that should be acknowledged. The method, like other time-domain
approaches, is sensitive to standard audio processing operations such as compression or filtering.
These operations may distort or erase the embedded data, compromising the integrity of the
hidden message. In addition, the fixed nature of the EOF marker may become a weakness if an
attacker is familiar with the pattern, as it could make the embedded data easier to detect or
manipulate. These aspects point to possible future improvements, such as adopting a more
dynamic EOF mechanism or integrating basic encryption to enhance security.
6. CONCLUSION
This study has explored the use of audio steganography through a modified Least Significant Bit
(LSB) method to securely embed confidential data within audio signals. Traditional LSB
techniques, while widely used for their simplicity and imperceptibility, are often limited by their
embedding capacity and vulnerability to signal manipulation. To address these shortcomings, a
custom binary encoding scheme and an End-of-File (EOF) marker were introduced in this work,
aiming to improve both the efficiency and accuracy of the data embedding and extraction
processes. The proposed technique demonstrated several advantages. First, it maintains the
perceptual quality of the audio, as confirmed by waveform and spectrogram analyses that showed
no significant distortion. Second, the incorporation of a fixed EOF marker helped enhance the
reliability of data retrieval by clearly signalling the end of the embedded message, thereby
reducing decoding errors.
However, certain limitations must be acknowledged. The technique remains susceptible to
common audio processing operations such as compression, filtering, or resampling, which could
compromise the integrity of the hidden data. Additionally, the fixed nature of the EOF marker,
while effective in controlled scenarios, may be predictable in adversarial contexts. Future work
may focus on enhancing robustness by integrating lightweight encryption or adaptive EOF
schemes that vary dynamically to avoid pattern detection. Exploration of hybrid methods
combining time- and frequency-domain embedding could further improve resistance to signal
degradation. Overall, this research contributes a practical, efficient advancement to the field of

38
audio steganography and opens the door to more secure and scalable methods for covert
communication.
REFERENCES
[1] J. Kunhoth, N. Subramanian, S. Al‐Maadeed, and A. Bouridane, “Video steganography: recent
advances and challenges,” Apr. 04, 2023, Springer Science+Business Media. doi: 10.1007/s11042-
023-14844-w.
[2] S. D. M. Satar, N. A. Hamid, F. Ghazali, R. Muda, and M. Mamat, “A New Model for Hiding Text
in an Image Using Logical Connective,” Jun. 30, 2015, Global Vision Press. doi:
10.14257/ijmue.2015.10.6.20.
[3] A.H. Lone, A. W. Lone and M. Uddin, “A novel scheme for image authentication and secret data
sharing,” 2016. https://doi.org/10.5815/ijcnis.2016.09.02.
[4] T. N. Aruna, L. R. Nandika, C. I. Sneha, T. J. Xavier, and T. M. George, “Text, Image and Audio
Steganography.” Apr. 2023.
[5] R. Shanthakumari, E. M. R. Devi, R. Rajadevi, and B. Bharaneeshwar, “Information Hiding in
Audio Steganography using LSB Matching Revisited,” May 01, 2021, IOP Publishing. doi:
10.1088/1742-6596/1911/1/012027.
[6] M. B. Begum and Y. Venkataramani, “LSB Based Audio Steganography Based On Text
Compression,” Jan. 01, 2012, Elsevier BV. doi: 10.1016/j.proeng.2012.01.917.
[7] F. Aslantaş and C. Hanilçi, “Comparative Analysis Of Audio Steganography Methods,” Jan. 22,
2022. doi: 10.38088/jise.932549.
[8] S. Dussan, A. S. Henao, M. M. Silva, and S. A. Donado, “Audio Steganography Algorithm Based
On Phase Coding,” Nov. 03, 2023, Research Square (United States). doi: 10.21203/rs.3.rs-
3504599/v1.
[9] D. Verma and P. Singh, “Performance Analysis of Audio Steganography using Modified LSB
Technique.” Jan. 2017.
[10] K. P. Adhiya and S. A. Patil, “Hiding Text in Audio Using LSB Based Steganography,” Jun. 08,
2012. Accessed: Oct. 2024. [Online]. Available: http://pakacademicsearch.com/pdf-files/eng/510/8-
14%20Vol%202,%20No%203%20(2012).pdf.
[11] E. W. Abood et al., “Audio steganography with enhanced LSB method for securing encrypted text
with bit cycling,” Jan. 22, 2022, Institute of Advanced Engineering and Science (IAES). doi:
10.11591/eei.v11i1.3279.
[12] M. H. Sayed and T. M. Wahbi, “Information Security for Audio Steganography Using a Phase
Coding Method,” Jan. 01, 2024. doi: 10.59324/ejtas.2024.2(1).55.
AUTHORS
Nazirah Abd Hamid is a senior lecturer in University Sultan Zainal Abidin,
Terengganu, Malaysia. She holds a degree in Bachelor of Information Technology
from University Utara Malaysia (UUM), in 2004, M. Sc. Com. (Information Security)
from University Teknologi Malaysia (UTM), in 2011 and PhD in Computer Science
from Universiti Teknikal Malaysia Melaka (UTeM), in 2023. Her research interests are
Information Security, Cyber Security, Pattern Recognition and Data Mining.
Siti Dhalila Mohd Satar received her Ph.D. from Universiti Putra Malaysia in 2024
and her M.Sc. from Universiti Teknologi Malaysia in 2012. She is a Lecturer at
Universiti Sultan Zainal Abidin, specializing in access control security systems, security
services—including digital forensics, steganography, network security, and
biometrics—and data security. Her research contributions focus on advancing security
mechanisms to protect digital ecosystems.

39
Mohd Fadzil Abdul Kadir received a Ph.D. in engineering (system engineering) from
the Mie University, Mie, Japan, in 2012. Since 2006, he has been with the Faculty of
Informatics and Computing, Universiti Sultan Zainal Abidin, where he is currently a
Senior Lecturer. His main areas of research interest are digital image processing,
pattern recognition, information security, and cryptography. He is also a member of the
Malaysia Board of Technologists. He can be contacted at email: fadzil@unisza.edu.my.
Ahmad Faisal Amri Abidin @ Bharun received his M. Sc. Com from Universiti Putra
Malaysia in 2008. He is a senior lecturer specializing in Security Services (Digital
Forensic, Steganography, Network Security, Public Key Infrastructure and Biometrics).
He has almost 15 years of experience in cybersecurity. In addition, he has led various
cybersecurity workshops.
Mohamad Afendee Mohamed received his Ph.D. in Mathematical Cryptography
from Universiti Putra Malaysia in 2011. Upon completion, he served the university for
three years as a senior lecturer. In 2014, he moved to Universiti Sultan Zainal Abidin
and later assumed an associate professor position. His current research interests
include both theoretical and application issues in the domain of data security and
mobile and wireless networking. He has authored more than 100 articles that have
appeared in various journals, book chapters, and conference proceedings. He can be
contacted at email:
Mumtazimah Mohamad was born in Terengganu, Malaysia. She received a bachelor’s
degree in information technology from Universiti Kebangsaan Malaysia, in 2000, an
M.Sc. degree in computer science from Universiti Putra Malaysia, and a Ph.D. degree in
computer science from Universiti Malaysia Terengganu in 2014. She was a Junior
Lecturer in 2000. Currently, she is an Associate Professor with the Department of
Computer Science, Faculty of Informatics and Computing (FIK), Universiti Sultan
Zainal Abidin, Terengganu, Malaysia. She has published over 50 research articles in
peer-reviewed journals, book chapters, and proceedings. She has been appointed as a reviewer and
technical committee for numerous conferences and journals and has worked as a researcher in several
nationally funded Research and Development projects. Her research interests include pattern recognition,
machine learning, artificial intelligence, and parallel processing.
Muhammad Izzul Haziq Bin Rahim is Final Year Student in University Sultan
Zainal Abidin, Terengganu, Malaysia. He currently pursue his study in B.Sc
Computer Network Security focusing on Network Architecture, Cyber Defence,
Cyber Attack and Digital Forensic. He also have experience with IoT device while
participate in many trainee workshops and involvement in R&D related to IoT
projects.

AUDIO STEGANOGRAPHY USING LEAST SIGNIFICANT BIT METHOD FOR CONFIDENTIAL DATA EMBEDDING

More Related Content

Similar to AUDIO STEGANOGRAPHY USING LEAST SIGNIFICANT BIT METHOD FOR CONFIDENTIAL DATA EMBEDDING (20)

More from johnmathew9417 (10)

Recently uploaded (20)

AUDIO STEGANOGRAPHY USING LEAST SIGNIFICANT BIT METHOD FOR CONFIDENTIAL DATA EMBEDDING