Localized & self adaptive audio watermarking algorithm in the wavelet domain
Introduction
Watermarking of audio signals have been in existence since last few years for secure transmission of audio signals over the network addressing copyright issues, piracy issues, ownership issues, fingerprinting and military applications etc. There are wide range of audio watermarking algorithms that are available in the existing literature for varied applications. Audio watermarking can be done in time domain, frequency domain, wavelet domain, cepstrum domain etc. Audio watermarking algorithms are basically divided into two categories: blind and non-blind. If original signal or watermark is required at the receiver side, then the watermarking algorithm is said to be non-blind, else it is blind algorithm. It is very challenging to design a blind audio watermarking algorithm because detection of efficient watermark at the receiver end, after application of various signal processing attacks is a crucial task. The embedding algorithm should be effective enough such that the original signal and the watermarked signal are perceptually similar. The watermarked signal must withstand serious signal processing attacks in order to extract the exact watermark with good accuracy. Imperceptibility, robustness and payload are the three contradictory design requirements of an audio watermarking algorithm [1]. In order to design an efficient audio watermarking algorithm, the three contradictory design parameters of the watermarking algorithm should be optimized. There are a large number of existing algorithms which provide a good watermarking solution at a high payload with good robustness against signal processing attacks maintaining perceptual constraints [2], [3], [4], [5], [6]. However, these algorithms may be suitable for some particular set of audio signals because property of audio signals is not taken into consideration while designing the embedding and extraction algorithm. In order to provide a solution to this problem, many researchers have proposed adaptive audio watermarking algorithm in which embedding strength of the algorithm is decided by using some of the audio features. The embedding strength here represents quantization parameter or some alpha factor etc. Li et al. has presented an adaptive audio watermarking algorithm in wavelet domain based on SNR to determine watermark embedding intensity by using scaling parameter,α [7]. Pooyan et al. [8] proposed a robust method of audio watermarking in wavelet domain which determines the quantization and embedding strength adaptively according to the characteristics of human auditory system (HAS). Wang et al. presented an adaptive audio watermarking algorithm in discrete multi wavelet transformation based on the energy relationship of two consecutive frames in multi wavelet domain [9]. Dahui Li presented an adaptive audio watermarking algorithm in DCT and DWT domain. This algorithm also chooses the quantization step according to the masking properties of HAS [10]. Dutta et al. proposed an efficient watermarking algorithm which embeds watermark data adaptively in SVD and DWT domain. In [11] high energy peaks are used to evaluate degree of embedding for each audio frame. Youssef presented a novel hybrid fuzzy self-adaptive digital audio watermarking scheme (HFSA-A W) in DWT which is based on local audio features [12]. Fuzzy c-means clustering is used in [12] for segmentation of audio features related to rhythm, timbre and harmony to estimate the strength of a frame for each sub-band and to ensure that the embedded watermark in the original audio is self-adaptive. But the method is non-blind because it requires an original signal at the receiver end for watermark extraction. Peng et al. presented an adaptive blind audio watermarking algorithm in wavelet domain based on local audio feature and support vector regression (SVR) [13]. In their paper, frame energy and maximal peaks of its all sub-bands are extracted as local features and SVR is used to model the relationship between the local features and the embedding strength of the audio frame in order to adaptively control the embedding strength of the audio frame.
It can be seen from the state-of-art that there exists various adaptive algorithms addressing the issue of robustness and imperceptibility in audio watermarking. In these algorithms, watermark is made inaudible by using different embedding intensities in different audio frames that depends on some features of that particular audio frame. Robustness is another design parameter of audio watermarking which is optimized with good perceptual transparency. Watermark payload capacity of the signal is not being considered in these methods. Different audio signals may have different capacity of carrying watermark data. For example, if an audio signal has low energy or less noise but carrying more watermarking bits than its capacity for fixed payload watermarking algorithm, it will lead to overloading of audio signal with watermark data. The changes due to insertion of watermark may be audible to human ear or it may also affect the robustness of the algorithm. In other case, if a high energy audio signal carry less number of bits than its capacity, this will lead to under loading of audio signal with watermark data. In this case robustness or imperceptibility will not be disturbed or even may be improved. Thus, there is an unused room available which should be utilized for hiding watermarking bits. In order to overcome the issue of overloading and under loading of audio signal with watermark data, an algorithm is proposed in this paper by considering the features of audio signal such that payload capacity directly depends on the signal leading to an adaptive nature of watermarking algorithm. This paper presents an adaptive audio watermarking algorithm in wavelet domain with varied payload for different audio signals. This proposed algorithm can estimate the payload for the audio signal by using local audio features like energy, zero cross mean etc.
The main contribution of this paper is an adaptive audio watermarking algorithm that can estimate the watermark payload by using local audio features from the host audio signal. This will optimize the payload that to an acceptable limit for an audio signal under perceptual transparency constraints. Unlike various existing algorithms that involves uniform payload, this adaptive algorithm will help in resolving the issue of over-loading and under-loading the audio signal with the watermark data. This will ensure that the most optimized payload is employed for each host audio signal depending on its local properties which makes the algorithm adaptive in nature. To address the design requirements like imperceptibility, robustness and adaptive payload which are mutually contradictory, the watermarking method is strategically chosen based on a wavelet decomposition method.
The rest of the paper is organized as follows: Section 2 describes the challenges faced while designing an adaptive audio watermarking algorithm, detailed methodology of the proposed algorithm is given in Section 3, experimental results are discussed in Section 4 and finally Section 5 concludes the paper and proposes a future prospect.
Section snippets
Adaptive audio watermarking: challenges
In any watermarking method, payload may affect both imperceptibility and robustness of the algorithm. Generally, more payload has less robustness and lower imperceptibility, whereas watermarking algorithms with less payload rates can retain watermark even under signal processing attacks exhibiting good imperceptibility and robustness of the algorithm. In general, existing audio watermarking techniques have fixed watermarking payload for any type of signal without considering the nature of the
Proposed methodology
This paper presents a localized audio feature based adaptive audio watermarking algorithm in wavelet domain having good robustness against signal processing attacks without disturbing the audible quality of the watermarked signal at different payload rates for different types of signals. In this paper, payload for an audio signal is found by evaluating the different features of the framed audio signal. This section describes the detailed methodology used for design and implementation of the
Experimental results
During experimentation, different types of audio samples are used to check the efficiency of the proposed algorithm against various parameters of audio watermarking algorithms. In the experimentation, Δ and ϒare considered as 0.08 and 0.06 respectively.
Some of the audio signals which are used in experimental setup are listed in Table 3. The database used for the experimentation of the proposed algorithm consists of approximately 115 audio files. All audio files used for experimentation are of
Conclusion
In this paper, an adaptive audio watermarking algorithm is proposed, which is designed to evaluate payload of audio signal on the basis of localized audio features. The proposed adaptive algorithm aims to resolve the problem of over-loading and under-loading the audio signals with watermark data making the payload optimized for each individual audio host signal. PCA is used in this paper for finding the discriminatory features which has relevance with the payload of the audio signal. A
Kaur A. received B.Tech & M.Tech in CSE from PTU, Jalandhar, India in 2006 and 2008 respectively. She is currently pursuing Ph.D. in CSE at Amity University, Noida, India.
Since 2008, she is with CSE department, Amity University. Her research interest includes audio watermarking, software reliability and image processing. She has published various peer reviewed papers in International Conferences and journals.
References (27)
- et al.
Perceptual based DWPT-DCT framework for selective blind audio watermarking
(2014) - et al.
Techniques for data hiding
IBM Syst J
(1996) - et al.
Time-spread echo-based audio watermarking with optimized imperceptibility and robustness
IEEE/ACM Trans Audio Speech Lang Process
(2015) - et al.
A perceptible watermarking algorithm for audio signals”
Multimedia tools & applications
(2014) - et al.
Robust audio watermarking using improved TS echo hiding
(2009) - et al.
A blind audio watermarking algorithm robust against synchronization attacks
- et al.
A Robust watermarking algorithm based on multi resolution decomposition of audio signal
- et al.
Adaptive audio watermarking algorithm based on SNR in wavelet domain
- et al.
Adaptive and robust audio watermarking in wavelet domain
- et al.
A robust watermarking scheme based on audio features in multiwavelet domain
A new algorithm of self-adaptability for digital audio watermark with energy quantization
An adaptive robust watermarking algorithm for audio signals using SVD
Transactions on computational science
HFSA-AW: a hybrid fuzzy self-adaptive audio watermarking
Cited by (24)
Robust audio watermarking scheme based on fractional Charlier moment transform and dual tree complex wavelet transform
2022, Expert Systems with ApplicationsCitation Excerpt :In the recent past, audio watermarking methods have been proposed. They are classified into two major types based on the domain where the watermark has been embedded: time-domain methods (Bassia et al., 2001), (Lie & Chang, 2006), (Wang et al., 2008) and transform-domain methods (Wang et al., 2013), (Hu et al., 2014), (Hu et al., 2017), (Kaur et al., 2017), (Karajeh et al., 2019), (Erçelebi & Batakçı, 2009), (Dhar, 2014), (Hu & Hsu, 2015), (Bhat et al., 2011a), (Ogura et al., 2017). Time-domain methods are the simplest because they are based on direct modification of host audio samples.
Efficient watermarking algorithm for digital audio/speech signal
2022, Digital Signal Processing: A Review JournalCitation Excerpt :Liu et al. [21] defined and discussed the feature coefficients cross-correlation degree (CCCD) of speech signal, and then explored a watermark embedding method based on the feature, in order to enlarge the embedding capacity and solve the security issue of watermark schemes based on public features. Based on localized features, Kaur et al. [22] proposed an adaptive audio watermarking algorithm where the capacity of this algorithm is adaptive and dependent on the nature of the watermarked audio signal. Saadi et al. [9] proposed a blind watermarking algorithm of speech and audio signals based on DWT and DCT transformations and based on sub-sampling as decomposition technique for correlation purpose.
Digital audio steganography: Systematic review, classification, and analysis of the current state of the art
2020, Computer Science ReviewCitation Excerpt :Cryptography modifies a secret message into a meaningless and unreadable form to deny attackers from understanding it, whereas steganography confidentially conceals secret messages and their existence in an innocent carrier. Perceptual transparency, hiding capacity, and robustness are the main requirements of steganography techniques used for evaluating the performance of any steganography approach (Fig. 1) [1]. Perceptual transparency is related to the similarity between the original cover and the stego files; it represents the degradation level of the cover files after hiding a secret message [2].
Novel secured scheme for blind audio/speech norm-space watermarking by Arnold algorithm
2019, Signal ProcessingCitation Excerpt :The watermark embedding is performed in [4] by modulating the vectors in the DCT domain subject to an auditory masking constraint and the abrupt artefacts in frame boundaries are further rectified via linear interpolation over transition areas. Paper [5] presents an adaptive blind audio watermarking algorithm in the wavelet domain to optimize the payload under the perceptual transparency constraints of audio signal by strategically using some of its local features. A blind and robust audio watermarking scheme based on SVD–DCT with the chaotic synchronization code technique is given in [6] by embedding a binary watermark into the high-frequency band of the SVD–DCT block blindly.
Optimised Cover Selection for Audio Steganography Using Multi-Objective Evolutionary Algorithm
2023, Journal of Information and Communication TechnologyA semi fragile watermarking algorithm based on compressed sensing applied for audio tampering detection and recovery
2022, Multimedia Tools and Applications
Kaur A. received B.Tech & M.Tech in CSE from PTU, Jalandhar, India in 2006 and 2008 respectively. She is currently pursuing Ph.D. in CSE at Amity University, Noida, India.
Since 2008, she is with CSE department, Amity University. Her research interest includes audio watermarking, software reliability and image processing. She has published various peer reviewed papers in International Conferences and journals.
M.K. Dutta received M.Tech in ECE from Central University, Tezpur India. He did his PhD in Multimedia Data security and Signal Processing from UPTU.
Since 2010, he is associated with the ECE department, AUUP, India. His interests include data security, watermarking and steganography. He has published many peer-reviewed papers in International journals and conferences. He has filed 3 patents and is principal investigator of various funded projects.
Dr. K.M Soni obtained his B.E. (Electrical Engineering) and M.E. (Control and Instrumentation) degrees from Motilal Nehru Regional Engineering College (Presently known as Motilal Nehru National institute of Technology) Allahabad (U.P) and his Ph. D. from Jamia Millia Islamia (a Central University) New Delhi. He is currently associated with Amity School of Engineering and Technology, Amity University Uttar Pradesh, Noida (U.P) and has been involved with teaching the students in the area of Circuits / Networks and Systems, Signal and Systems, Digital Signal Processing, Control Systems, Modern/Digital Control Systems, Process Control, Power Electronics, Electrical Science and Machines etc. at undergraduate and postgraduate levels. He has also authored the books on Signals and Systems, Basic System Analysis, Network Analysis and Synthesis, and Advanced Control Systems. He has also supervised many B. Tech and M. Tech projects and presented/published technical papers in International Conferences and Journals. He is Life Member of Indian Society for Technical Education (ISTE), Institution for Electronics and Telecommunication Engineers (IETE), International Journal of Engineering Research and Industrial Applications (IJERIA), Member of IEEE, IET(U.K), CSI and other similar Organization.
Nidhi Taneja is associated with the Department of Electronics and Communication Engineering, Indira Gandhi Delhi Technological University for Women, DELHI. She did her Ph.D. from IIT, Roorkee in the field of Security solutions for still Visual Data. She has published many peer-reviewed research papers in International journals and conferences.