Você está na página 1de 6

International Conference on Computer and Communication Engineering (ICCCE 2012), 3-5 July 2012, Kuala Lumpur, Malaysia

Digital Audio Watermarking; Techniques and


Applications
R. F. Olanrewaju1, Othman Khalifa
Department of Electrical & Computer Engineering, Faculty of Engineering,
International Islamic University Malaysia
Kuala Lumpur Malaysia,
1
frashidah@yahoo.com

Abstract— Counterfeiting and piracy of intellectual properties are duplication, and content protection in digital audio has become
often pointed to as an indispensable problems across the globe, more apparent and sensitive issues.
though more intense in some region than others. Currently over Recently, Motion Picture Association of America (MPAA)
millions of digital audio files like digital songs are copied illegally is on a world tour trying to convince politicians to introduce
during file-sharing over the networks, costing loss of revenue to the
music and broadcasting industries. Right owners such artists, legislation that criminalizes the recording of movies in
composers, and producers have been searching for ways to protect cinemas. The MPAA already succeeded in Canada, Japan and
their creations, discoveries and inventions. These challenges Italy, and their next stop is the UK [2]. Another breakthrough
prompted significant research in Digital Watermarking (DW) to coalition of the MPAA, Recording Industry Association of
develop efficient methods to protect copyright and authentication America (RIAA) and other copyright holders have signed an
messages in digital audio in order to prevent forgery and agreement with AT&T, Cablevision, Comcast, Time Warner
impersonation. This paper brings to view techniques and applications
of digital audio watermarking signals. Requirements and Methods of
Cable and Verizon to curb piracy. Under the agreement, the
audio watermark were indentified and classified accordingly. A new ISPs agree to send “copyright alerts” to subscribers whose
intelligent audio watermarking technique in a combine domain is Internet connections are used for copyright infringement.
proposed to complement the current methods. It is expected that this Repeated offenders will not be disconnected from the Internet,
paper will serve as a launch pad for researchers interested in audio but could be slowed down instead [3].
watermarking. These challenges prompted significant research to develop
efficient methods to protect copyright and authentication
Keywords- digital watermarking; audio signals; copyright
messages in digital audios in order to prevent forgery and
protection; watermark attacks
impersonation [4]
Digital Audio Watermarking (DAW) is inaudibly altering of
audio signal to embed message or marks about the host itself
which can be detected or extracted afterwards to make an
I. INTRODUCTION assertion about the original audio signal. In digital audio
Happenings in the music industry are drastically new watermarking, the original audio being communicated is the
phenomenon and a natural development. The new cover or host signal and watermark provides additional
phenomenon is the fact that the Internet is changing the information about the cover. Digital audio files such as songs
storage, retrieval and distribution model for songs. Access to are the most susceptible to copyright infringements because of
broadband transmission as well as digitization of audio signals conditions attached to it [5]. For example, most digital songs
is the main success of this change. Digital representations of are offered online in advance to be downloaded as
copyrighted audio files such as radio programs, music promotional copies. These promotional copies can be used by
telephone calls, phone banking, emergency calls, air traffic pirates.
communication and recordings of cultural heritage Digital watermarking can be effectively used to limit the scope
preservations offer several distinct opportunity and advantages of counterfeiting and piracy of audio files. This is due to
over the analog counterpart. Thus editing is easy because one crucial features such as imperceptibility, inseparability of the
can access the exact discrete locations that should be changed content from the watermark, and intrinsic ability to undergo
and copying is simple with no loss of fidelity. Digital audios same transformation experienced by the host signal, which
are easily transmitted over networked information systems digital watermark possesses. This preference has been
faster and inexpensively [1]. This development results from established to provide improved security [6].
unrestricted access offered by the digital technology Research reports in Digital Audio Watermarking is relatively
particularly large-scale unauthorized copying, downloading, low in percentage compare to image and video watermarking
and distribution of multimedia work. Consequently, the due to sensitivity of Human Auditory System (HAS). DAW
importance of authentication, content verification, authorized scheme rely on the imperfection of HAS, which is more

978-1-4673-0479-5/12/$31.00 ©2012 IEEE

830
sensitive than other sensory motors. Hence, designing of a includes processing properties such as, payload, security,
balanced auditory scheme is a complex issue. Furthermore, the complexity of the watermarking algorithm, perceptual
amount of information that can be embedded into audio signal transparency, robustness and application specific requirements
is lesser compared to image or in video. This is because audio [7, 9]. Robustness of the algorithm to attacks and quality of
is a one dimensional signal. At present, publications in audio the watermarked audio tracks are related properties that are
watermarking have steadily increased since year 2000[7, 8]. indispensable. All applications presupposing protection and
Thus digital audio watermarking domain has potential to be use in verification of the watermarking systems require this
used as a system for managing Intellectual Property rights, and type of marking in order to survive any kind of alterations or
can be used not only to know the author of a particular audio intentional removal introduced by standard or malicious
file, but also record the path a particular file takes if it is processing and attacks [10]. Figure 1 shows some common
distributed in an unauthorized manner [5]. attacks on audio watermarking system.

II. REQUIREMENTS AND TECHNIQUES OF DIGITAL AUDIO


WATERMARKING DESIGN

A. Digital Audio Watermarking Requirements


To design an optimal audio watermarking system, some
design features need to be taken into consideration. This

Figure 1: Classification of digital audio watermarking attacks

original signal known as non blind/informed detection or


without the original audio signal called a blind detection.
B. Digital Audio Watermarking System Modules
A typical audio digital watermarking system consists of C. Audio Watermarking Techniques and Algorithms
two generic modules, the watermark embedding module and The two main streams of watermark embedding are to
watermark recovery module also known as extractor. The embed either in time or transform domain as shown in Figure
embedding module inserts the watermark information in the 3.
audio signal while the recovery module extracts /detects the Though recently, reserchers have been using combined
watermarked information as illustrated in Figure 2. In some domain to improve on the robutness of audio watermarking
systems, the detection can be with the availability of the algorithms..

831
Figure 2: Digital watermarking scheme with embedding and extracting module

into series of short sequence length. The watermark is then


embedded as phase shifts in the phase spectrum into the first
block of the audio stream. This leads to inaudible encoding in
terms of Signal-to- Perceived Noise Ratio (SPNR) and the
watermark gets camouflaged in the audio signal[18].The
major drawback of phase coding is low payload, because,
watermark can be embedded into the first block only. This
also implies that the watermark is localized which can be
easily remove by attack like cropping. Furthermore, it is a non
blind and it applicability is limited [7].
To overcome the weakness in phase coding, phase modulation
[19, 20], a time-varying of all-pass filter is modulated
individually in the audio signal. Each coefficient is
independently and inaudibly altered. A disadvantage of phase
Figure 3: Digital Watermarking Domains coding is the complexity of this method due to the number of
transformations which have to be computed for detection, that
D. Time Domain Audio Watermarking is, performed in the Cepstrum domain [21]. It is also a non
Time domain watermarking was the first watermarking blind algorithm.
schemes that were investigated by researchers. Embedding in Some algorithms [22, 23] in transform domain are based on
the temporal domain means that the watermark is embedded Echo hiding techniques. Echo hiding algorithms embed
directly into the host audio by modifying the arrays or adding watermarks into a signal by adding echos to produce a marked
a pseudorandom noise pattern to the audio data [11]. Temporal signal.
domain techniques include Low bit coding such as Least The echo-hiding watermarking is modeled as;
Significant Bit coding (LSB). This is easily achieved by the
camouflage of noises inherent to digital signal acquisition (1)
process [10, 11, 12]. It requires altering the LSB of individual
sampled of the digitized audio stream. Though this technique Where y(n) is the watermarked signal segment, x(n) is a
is easy to design and can embed large data, however it is segment of the audio signal, h(n) s the echo kernel, and the *
vulnerable to even a slight audio distortion, such as simple file operator denotes convolution.
conversion. It also causes audio degradation because of Here, three parameters are considered; the amplitude, delay
random changes especially in LSB. time and decay rate of the host signal. The echo is not easily
resolved as all the three parameters are set below the human
E. Transform Domain Audio watermarking audible threshold limit [24]. Echo hiding is a blind method
Audio watermarking techniques that work in frequency that allows for a higher payload and provides superior
domain, take the advantage of audio masking characteristics of robustness when compared to the noise inducing methods.
Human Auditory System, (HAS) to embed an inaudible Spread Spectrum (SS) coding method [25, 26], a technique
watermark signal in digital audio signals. Transforming audio that randomly spreads the bits of the secret data message
signal from time domain to frequency domain enables across the frequency spectrum of the audio signal [27]. SS
watermarking system to embed the watermark into require that the transmitter and receiver are synchronized. It
perceptually significant components. This provides the system uses a separate code independent of the host signal to encode
with a high level of robustness and inaudibleness [15]. and decode the watermark, Though SS detection does not
Techniques such as Phase coding [16][17] uses phase shift require the original recording and it is robust to attacks
in phase spectrum of signal carrier to split the carrier signal compared to the phase coding and phase modulation, however
it also introduces noise.

832
Mathematically give as; Psychoacoustic Masking and Bitstream method [39],
Frequency Domain Shaping and Time Domain Weighting
(2) [40], et.c. Through combine techniques, there is a better
assurance that the watermark will not be audible under all
where M(n) is the sequence of data from of the original audio, bitrates of compression technology. Additional, it has low
X(n) is the watermark sequence, λ is a gain factor controlling computational complexity, thus is ideal for transactional, or
the embedding strength, and M‘(n) is the sequence of per sale, watermarking. It also does not degrade the audio
watermarked audio. quality. [41].
Patchwork [28, 29, 30], is a statistical technique that uses
redundant pattern encoding to embed a message in an audio
signal. It mainly involve two steps:
III. DIGITAL AUDIO WATERMARKING APPLICATIONS
1. choose two patches pseudo-randomly and
2. add the small constant value to the sample values of Copy prevention and copyright protection applications have
one patch X and subtract the same value from the been the brain behind audio watermarking. However, it’s not
sample values of another patch Y [31]. limited to that. Applications such as broadcast monitoring,
Patchwork is modeled as; content authentication and fingerprinting have benefited a lot
from audio watermarking. These applications are increasing
(3) rapidly. Recently, digital audio watermarking has taken a new
dimension. Nowadays, watermarks are use to stop music
(4)
writers from leaking advance promotional copies onto the
internet [37].
where and are sample values of the patchwork sets and X In addition, watermark has been used to stop camcording
and Y respectively. Though 3DWFKZRUN DOJRULWKP GRHV pirates which poses a serious threat to the profits of music
QRW UHTXLUH WKH RULJLQDO KRVW VLJQDO LQ WKH industries. In order to stop the cammers in their tracks, theater
SURFHVVRIZDWHUPDUNGHWHFWLRQ EOLQGZDWHUPDUNLQJ owners have been equipped with night-vision goggles. In this
GHWHFWLRQ +RZHYHUWhe performance of the patchwork case, watermarks are use locate camcording pirates seat. For
algorithm depends on the distance between two sample means example, the night vision goggles helped to spot “Batman and
and which affects the inaudibility. Furthermore, the patchwork Bond” pirates. Besides, a teenager was caught for recording a
algorithm has originally been designed for images, therefore, few seconds of the movie “Transformers” on her cell-phone
it is susceptible to potentially reversed engineering by players [42].
for removal. In some studios, watermarks are recorded in the soundtrack of
theatrical releases, and when a pirated recording appears it
will be possible to determine the place and date of its creation.
F. Compressed Domain Audio Watermarking Such watermark will confirm the changes that have occurred.
Compression capability of the MPEG standard [32] has Furthermore, authors of musical composition embed a
established a preferred coding technique for audiovisual watermark by using specialized software and distribute the
content [33]. This development, coupled with the fact that in product. If a dispute about the authorship arises, watermark
compress domain [34], watermark signal are embedded will serve as a means of litigation. Such watermark will allow
directly into the MPEG audio bit stream rather than encoding tracing the path of illegal distribution of audio files as well.
and decoding in uncompress domain [35, 36] is of advantage. In recent times, watermarks are embedded as a function in
This implies that there is no need to change the original telephoning system which will further confirm the identity of
bitstream syntax. This is due to the fact that MPEG audio the called and the caller [2].
compression is a lossy algorithm and uses the special nature of .
the HAS because the dynamic range of the auditory system
decreases with decreasing frequency. It removes the
perceptually irrelevant parts of the audio and makes the audio IV. FUTURE DIRECTION OF DIGITAL AUDIO WATERMARKING
signal distortion inaudible to human ear. Such systems are
suitable for “pay audio” scenario, where the provider stores In the nearest future, watermark will be equipped with
audio contents in compressed format [37]. During download intelligence that can potentially tell the content of an audio
of music, the customer identifies himself/herself with his/her file, where it originated from, the distribution channel, date
unique customer ID, which therefore is known to the provider distributed, to whom it was distributed (legal customer profile)
during delivery. and possibly the type of media the track was last used on. For
example the watermark content can go as follows;
G. Combined Domain Audio Watermarking
'this content is "Stairway to Heaven," distributed through
Combined domain audio watermarking is the utilizing of Starbucks, last transferred through the Hear Music kiosk in
multi-dimensional approach. This includes the use of downtown San Francisco on Dec. 8th, to a registered Hear
Psychoacoustic Models and Spread Spectrum [38],

833
Music user who has this profile and uses a Microsoft Zune." characteristics of Neural Network such as learning and
[41]. adaptive capabilities, pattern mapping and classification and
Such system can be easily implemented on connected devices, ability to generalize, not only to reproduce previously seen
like the iPhone, iPods etc. data, but also provide correct predictions in similar situations
gives the trained networks ability to recover the watermark
from the watermarked data [4]. The incorporation of such
V. WAY FORWARD consideration is referred to as “Intelligent Combined Domain
Audio Watermarking” (i-CAW)
Intelligent Combined Domain Audio Watermarking . In this newly propose technique, embedding watermark in
From the above reviews, several studies have been carried out such trained location in a combined domain will bring balance
in audio watermarking, ranging from temporal to combined among inaudibility, payload and robustness at a low
domain. It is noticed that there is no single method that fulfill computational complexity.
all the audio watermarking requirements due to a trade-off This suggests that investigating and researching in such niche
between robustness, inaudibility and payload. Fixing any area will lead to design of an attack proof watermarking
dimension such as robustness, there exist tradeoff scheme as well as solve the problem of audio degradation.
relationships between the other two dimensions. However, Schematic diagram of the proposed domain is as shown in
there is a need to explore the use of intelligent agent such as Figure 4.
artificial neural network (ANN) to bring balance among the
requirements as well as train locations in which the watermark
can be safely embedded. In such that the watermark is neither
destroyed nor be removed by attacks as well as does not
degrade the quality of the audio signal. Owing to the inherent

Figure 4: Schematic diagram of proposed Intelligent Combined Domain Audio Watermarking i-CAW
VI. CONCLUSION REFERENCES
Digital Audio Watermarking technology is now coming a vital [1] S. Sherekar, V. Thakare, and S. Jain, Role of Digital Watermark In E-
Governance and E-Commerce. IJCSNS, 8(1), pp. 257. 2008.
weapon to deter impersonification and piracy in the audio
industries. We have presented an overview of various audio [2] MPAA, RIAA Team Up With ISPs to ‘Alert’ Pirates,
<http://torrentfreak.com/mpaa-riaa-team-up-with-isps-to-curb-piracy-
watermarking schemes as well as the algorithmic 110707/
requirements. We discussed many of the current application of [3] http://speetech.com/press/analytics/the-use-of-audio-watermarking-in-
digital audio watering and it future application. There has not the-recording-industrymarking
been sufficient investigation into Intelligent digital audio [4] R. F. Olanrewaju, Development of An Intelligent Digital Watermarking
watermarking. In order for watermarking to be successful, we Algorithm Via Safe Region”, Unpublished PhD thesis, International
propose application of ANN in combined domain to balance Islamic University Malaysia, 2011.
digital audio watermarking requirements [5] M. O Agbaje , A.T Akinwale and A.N Njah, Audio Watermarking: A
Critical Review , International Journal of Scientific & Engineering
Research, 2(11), 2011.

834
[6] R. F. Olanrewaju, A. A. Aburas, and O.O. Khalifa and A. Abdalla [26] N. Meghanathan, and L. Nayak, A review of the audio and video
State-of-The-Art Application of Artificial Neural Network In Digital steganalysis algorithms, ACM, 2010.
Watermarking And The Way Forward. Paper Presented At The [27] P. Bassia, I. Pitas, and N. Nikolaidis, “Robust audio watermarking in the
International Conference on Computing & Informatics (ICOCI 09), time domain,” IEEE Trans. Multimedia, vol. 3, no. 2, pp. 232–241, Jun.
Kuala Lumpur Malaysia. Malaysia, 2009. 2001.
[7] M. K., M. Arnold, M. Schmucker and S. D. Wolthusen, Techniques [28] W. Bender, D. Gruhl, N. Morimoto, & A. Lu, Techniques For Data
and applications of digital watermarking and content protection, Artech Hiding. IBM Systems Journal, 35(3/4), PP. 313-336, 1996.
House Publishers, 2003. [29] N. K. Kalantari, M. A. Akhaee, S. M. Ahadi, and H. Amindavar, IEEE
Transactions on Audio, Speech, and Language Processing, 17(6), 2009.
[8] J. S. Pan, H. C. Huang, L. C. Jain, Intelligent Watermarking
Techniques, World Scientific Pub Co Inc. 2004. [30] I. K.Yeo and H. J. Kim, Modified patchwork algorithm:A novel audio
watermarking scheme, IEEE Trans. Speech Audio Process., 11(4), pp.
[9] N. Cvejic, and T. Seppanen. Digital audio watermarking techniques and
381–386, Jul. 2003.
technologies: applications and benchmarks, IGI Global. 2007.
[31] ISO/IEC 13818-2, Information technology-Generic coding of moving
[10] I. Cox,, M. Miller, et al. Digital watermarking and steganography,
pictures and associated audio: Video., 2000.
Morgan Kaufmann, 2008.
[32] D. Simitopoulos, Fast Watermarking of MPEG-1/2 Streams Using
[11] J. Dittmann, M. Steinebach, T. Kunkelmann, and L. Stoffels, H2O4M- Compressed-Domain Perceptual Embedding and a Generalized
Watermarking For Media: Classification, Quality Evaluation, Design Correlator Detector EURASIP Journal on Applied Signal Processing,
Improvements. Proceedings of ACM Workshops on Multimedia, pp. 2004(8), pp 1088-1106, 2004.
107-110, 2000. [33] M. A. T. Alsalami and M. M. Al-Akaidi, Digital Audio Watermarking:
[12] Z. Gao, and Y. Zheng, Quality constrained compression using DWT- Survey,Proc. 17th European Simulation multiconference, De Montfort
based image quality metric. Circuits and Systems for Video Technology, UK, pp. 1-14, 2003.
IEEE Transactions on, 18(7), pp. 910-922., 2008. [34] C. Neubauer, and J. Herre, Audio watermarking of MPEG-2 AAC bit
[13] R. Van Schyndel, A Tirkel, and C. Osborne, A Digital Watermark. streams, Preprints-Audio Engineering Society, 2000.
IEEE International Conference Image Processing, pp. 86-90, 1994. [35] P. Artameeyanant, Wavelet audio watermark robust against MPEG
[14] S. Meng, W. Cao, Y. Yan and S. Li , Bit Replacement Audio compression, Control Automation and Systems (ICCAS), Inter. Conf. on
Watermarking Using Stereo Signals Advanced in Information Sciences , vol., no., pp.1375-1378, 27-30 Oct. 2010.
and Service Sciences, 2(3), September 2010. [36] E. V. Buskirk, Are Digital Music Watermarks a Blessing or a Curse?,
[15] I. Cox, J. Kilian, F. Leighton, and T. Shamoon, Secure Spread Spectrum 2007.
Watermarking for Multimedia. IEEE Transactions on Image Processing, http://www.wired.com/entertainment/music/commentary/listeningpost/2
6(12), pp. 1673-1687, 2002. 007/08/listeningpost_0820
[16] X. M. Chen, G. Doërr, M. Arnold, P.G Baunm. (2011). Efficient [37] C. Neubauer, and J. Herre, Digital Watermarking and its Influence on
coherent phase quantization for audio watermarking, IEEE International Audio Quality. 105th AES Convention, San Fransisco, California.
Conference on Acoustics, Speech and Signal Processing (ICASSP), Preprint 4823, 1998
2011. [38] J. Lacy, R. Quackenbush, A. Reibman , D. Shur and J. Snyder, On
[17] X. Dong, M. F. Bocko, Z. Ignjatovic, Data hiding via phase Combining Watermarking with Perceptual Coding. ICASSP Seattle,
manipulation of audio signals, IEEE Inter. Conf. on .Acoustics, Speech, Washington. MMSP1.9, 1998.
and Signal Processing, Proceedings. (ICASSP '04). vol.5, no., pp. V- [39] L. Boney, A. H. Tewfik, K. N Hamdy 1996 Digital Watermarks for
377-80 , 17-21 May 2004. Audio Signals. EUSIPCO-96, VIII European Signal Proc. Conf., Trieste,
[18] R. Ansari, H. Malik, and A. Khokhar, Data-Hiding In Audio Using Italy. (Patent pending)
Frequency-Selective Phase Alteration, IEEE Int. Conf. on ,Acoustics, [40] R. Eklund, Audio Watermarking Techniques,
Speech, and Signal Processing, pp. 389-392 vol.5, 17-21 May 2004 http://www.musemagic.com/papers/watermark.html.
[41] Audio Watermarks Locate Camcording Pirates
[19] A. Takahashi,, R. Nishimura, et al. Multiple watermarks for stereo audio http://torrentfreak.com/audio-watermarks-locate-camcording-pirates-
signals using phase-modulation techniques, IEEE Transactions on 090304/
Signal Processing, 53(2): 806-815, 2005.
[42] MPAA lawmaker, MPAA Pushes Lawmakers to Criminalize Movie
Camming, http://torrentfreak.com/mpaa-to-criminalize-movie-camming-
[20] M. Arnold, M. Arnold, P. Baum and W. Voeßing A Phase Modulation
070905/
Audio Watermarking technique Information Hiding. S. Katzenbeisser
and A.-R. Sadeghi, Springer Berlin / Heidelberg. 5806: 102-116, 2009.
S.-K. Lee and Y.- S. Ho, Digital Audio Watermarking in the ceptrum
domain, IEEE Transactions on Consumer and Electronic, pp, 744- 750,
vol 6, no 3, 2000.
[21] Y. Xiang, D. Peng, I. Natgunanathan, and W. Zhou, Effective
Pseudonoise Sequence and Decoding Function for Imperceptibility and
Robustness Enhancement in Time-Spread Echo-Based Audio
Watermarking, IEEE Transactions on Multimedia, 13(1), February 2011.
[22] W. Bender, D. Gruhl, N. Morimoto, and A. Lu,”Techniques for data
hiding,” IBM Systems Journal, vol. 35, pp. 313-336, 1996.
[23] H.O. Oh, J. W. Seok, J. W. Hong, and D. H Youn,”New echo
embedding technique for robust and imperceptible audio watermarking,”
IEEE International Conference on Acoustics, Speech, and Signal
Processing, vol. 3, pp. 1341-1344, 2001.
[24] D. Kirovski, and H. S. Malvar, Spread-spectrum watermarking of audio
signals, IEEE Transactions on Signal Processing, 51(4): 1020-1033,
2003
[25] I. J. Cox,, J. Kilian, T. Leighton, and T. Shamoon, Secure spread
spectrum watermarking for images, audio and video, IEEE Proc. Int.
Conf. on Image Processing, 1996.

835

Você também pode gostar