Some Audio Steganography and Watermarking algorithms

1. AUDIO WATERMARKING AND
STEGANOGRAPHY
Anirudh Shekhawat
Manan Shah
Prateek Srivastava
Pratik Poddar
Guided by: Prof. Bernard Menezes

2. DIGITAL WATERMARKING
 Embedding perceptually transparent data in
digital media
 Watermark can be detected and retrieved by a
computer algorithm
 Applications include broadcast monitoring,
fingerprinting, copyright protection and
steganography

3. STEGANOGRAPHY
 Steganography literally means "secret writing”
 Hiding data in a digital media
 Avoids the suspicion and scrutiny an encrypted
message would arouse
 Applications include "covert communication”
 Examples from history - Invisible ink, Scalp
tattoo, Pinholes

4. APPLICATIONS

5. COPYRIGHT PROTECTION
 Ownership information can be embedded in the
media
 Presence of the watermark can be demonstrated
to prove ownership
 The watermark must survive compression

6. FINGERPRINTING
 Embedding a serial number in each copy of a
media before distribution
 Can be used to trace down the originator of a
particular illegal copy of media
 The watermark must be secure against attacks

7. BROADCAST MONITORING
 Programs and advertisements broadcasted can be
monitored by an automated system
 Illegal broadcasts can be identified by monitoring
satellite nodes
 High Bit rate and low complexity are required

8. COVERT COMMUNICATION
 Embedding data such that existence of a
watermark cannot be detected
 Aimed at protecting the sender and receiver
rather than the message
 Terrorists have been reported to hide messages
in images to communicate

9. CHARACTERISTICS OF A
WATERMARKING ALGORITHM

10. PERCEPTUAL TRANSPARENCY
 Primary requirement in watermarking
 Watermark should be imperceptible to human
auditory system
 Watermarking is more difficult for audio as
compared to images

11. WATERMARK BIT RATE
 Represented in bits per second (bps)
 Required rates vary across applications (0.5 bps
in copyright protection and 15 bps in broadcast
monitoring)
 Attainable values depend upon level of
compression of audio

12. ROBUSTNESS
 The ability of the watermark to survive common
signal processing manipulations
 Required against a predefined set of
manipulations
 Required in some applications (radio broadcast
monitoring) but not at all required in some
(tampering detection)

13. BLIND AND INFORMED DETECTION
 Informed: with access to original(host) audio
 Blind: without access to original audio
 Informed techniques are more secure
 Examples
 Blind: Tampering detection, Information Carrier
 Informed: Steganography

14. SECURITY
 Adversary must not be able to detect the
existence of embedded data
 Not be able to extract or modify the data without
the secret key
 In some cases the watermark is encrypted before
being embedded

15. ALGORITHMS AND DEMO

16. 1) LEAST SIGNIFICANT BIT ENCODING
 Watermark added in the least significant bit of
amplitude
 Easy to embed and retrieve
 High bit rate
 Low robustness

17. WATERMARKING, COMPRESSION &
REDUNDANCY
 Lossy compression destroys watermark.
 To make watermark robust against compression,
we need sufficient redundancy
 LSB encoding: Robustness vs Imperceptibility?
 LSB Watermarking in JPEG done 
 Possible for MP3?

18. 2) PHASE MODULATION
 Embed watermark by modulating phase in host
audio
 Robust against signal processing manipulations
 Extraction or detection of watermark needs
original audio
 Suitable for applications where security and
robustness are important, e.g. copyright
protection

19. 3) FREQUENCY DOMAIN STEGANOGRAPHY
 Shanon Sampling theorem
 Sampling rate for CD is 44.1KHz the highest
frequency is 18KHz
 Average peak frequency which a human can hear
is 18Khz
22 – 18 = 4KHz band goes unused

20. UNDERLYING PRINCIPLE
 Use the 18Khz - 22KHz frequency band to hide
the message
Message
signal
Base
signal
Fig : Combined Signal

21. MERITS AND DEMERITS
 Merits
 Longer message can be hidden in a given base
 Less likely to be affected by errors during
transmission
 Demerits
 Message signal has limited frequency range
 Low recovery quality

22. 4) ECHO HIDING
 Embeds data by introducing “echo” in the original
signal.
 Resilient to lossy data compression algorithms.

23. ECHO ENCODING AND DECODING
Encoding
Decoding

24. THE END.. QUESTIONS?

25. REFERENCES
 Juergen Seitz, Digital Watermarking for Digital Media, ISBN
159140519X, 2005, Information Resources Press, Arlington, VA,
USA
 Nedeljko Cvejic, Algorithms for audio watermarking and
steganography, ISBN 9514273834, 2004, Oulu University Press,
Oulu
 C.H. Yeh & C.J Kuo, Digital watermarking through quasi m-arrays,
Proceedings of the IEEE Workshop on Signal Processing Systems
1999, 456-461
 Guy Belloch, Introduction to Data Compression, Draft version,
Algorithms in the real world
 Kuo S, Johnston J, Turin W & Quackenbush S Covert audio-
watermarking using perceptually tuned signal independent
multiband phase modulation, IEEE International Conference on
Acoustics, Speech, and Signal Processing 2002, 1753-1756

26. REFERENCES
 Foo, S.W., Yeo, T.H., & Huang, D.Y. (2001). An adaptive audio
watermarking system. Proceedings of the IEEE Region 10
International Conference on Electrical and Electronic Technology,
509–513.
 Huang, D.Y., & Yeo, Y.H. (2002). Robust and inaudible multi-echo
audio watermarking. Proceedings of the IEEE Pacific-Rim Conference
on Multimedia, 615–622.
 Bender, W., Gruhl D. Echo Hiding, International Workshop on
Information Hiding, 1996.