WO2008149307A1

WO2008149307A1 - Watermark generation and embedding

Info

Publication number: WO2008149307A1
Application number: PCT/IB2008/052214
Authority: WO
Inventors: Aweke N. Lemma; Mehmet U. Celik; Stefan Katzenbeisser; Minne Van Der Veen
Original assignee: Koninklijke Philips Electronics N.V.
Priority date: 2007-06-08
Filing date: 2008-06-05
Publication date: 2008-12-11

Abstract

A method for generating a watermark signal and embedding the watermark signalin an information signal. The information signal comprises a plurality of scale factors (SF k ) each of which is associated with a range of frequencies. The method comprises generating (21) said watermark signal by combining first and second sequences of values, at least one of said first and second sequences of values being generated by shifting an initial sequence of values in a manner determined by a payload. At least one of said scale factors (SF k ) is modified (22) based upon said watermark signal. In an application of the invention, a bitstream is encrypted by modifying the scale factors at a server side. At the client side, the bitstream is decrypted by modifying the scale factors such that a watermark is left in the decrypted signal.

Description

Watermark generation and embedding

FIELD OF INVENTION

The present invention relates to a method and apparatus for generating a watermark and embedding the generated watermark in an information signal.

BACKGROUND

Unauthorized distribution of digital media, such a music and movie files, is a serious problem and one of considerable concern to media owners. It is important to ensure that media distribution is properly controlled so as to ensure that a media owner's income stream is not adversely affected. It has been proposed that watermark data should be embedded within digital media signals so as to mitigate the problem of unauthorized distribution. Such watermarks take a variety of forms. For example, playback-control watermarks may be used so as to restrict access to particular digital media signals to particular devices authorized to access those signals, and to prevent other devices from obtaining access to those signals. Forensic watermarking is a technique which is intended to allow digital media signals which are distributed in an unauthorized manner to be traced to a particular authorized user. In this way, authorized users who initiate unauthorized distribution can be identified and appropriate action can be taken. Forensic watermarking is implemented in such a way that each authorized user is only allowed to access a digital media signal in which a watermark unique to that user has been embedded. All copies of the digital media signal can be traced back to the appropriate authorized user using embedded watermark.

In watermarking systems difficulties arise between various competing goals. It is desirable to provide watermarks which are difficult to remove once embedded, and which are easily detectable. It is however also important to ensure that a signal in which a watermark is embedded does not have user-perceptible artefacts, that is where a watermark is embedded in an audio signal, a user should not be aware of the presence of the watermark during playback of the audio signal. Many prior art systems have difficulty in providing an acceptable balance between the competing goals. In some arrangements a watermark is embedded in a baseband signal. Such arrangements generally allow a watermark to be embedded in such a way that its removal is not readily possible. However such arrangements are often disadvantageous in that the embedding process is complex. In alternative arrangements a watermark is embedded in an encoded or bitstream signal. Such alternative arrangements are typically less complex, but are disadvantageous in that a watermark can be more easily removed. Thus each of these arrangements has disadvantages.

It is an object of embodiments of the present invention to obviate or mitigate at least some of the problems outlined above.

SUMMARY OF INVENTION

According to the present invention, there is provided, a method and apparatus for generating a watermark signal and embedding the watermark signal in an information signal. The information signal comprises a plurality of scale factors each of which is associated with a range of frequencies. The method comprises generating said watermark signal by combining first and second sequences of values, at least one of said first and second sequences of values being generated by shifting an initial sequence of values in a manner determined by a payload; and modifying at least one of said scale factors based upon said watermark signal. Creating a watermark by combining first and second sequences of values provides a robust watermark. It has been found to be particularly effective to embed such a watermark in an information signal by modification of scale factors in the information signal. The information signal may comprise a plurality of frames. Each of the frames may comprise said plurality of scale factors and scale factors in each of said frames may be modified based upon a respective part of said watermark signal.

The first sequence may be generated by combining (for example interleaving) a first initial sequence with a sequence comprising values of equal magnitude by opposing sign. The second sequence may be generated by combining (for example interleaving) a second initial sequence with a sequence comprising values of equal magnitude by opposing sign.

The first and second sequences may be combined by interleaving said first and second sequences.

A plurality of watermark signals may be generated by combining respective first and second sequences of values, at least one of said respective first and second sequences of values may be generated by shifting an initial sequence of values in a manner determined by a payload.

The method may further comprise modifying a plurality of said scale factors, each of said scale factors may be modified based upon a predetermined watermark signal. The information signal may be an encoded signal. The scale factors may be modified when said encoded signal is decoded. The scale factors may alternatively be modified during an encoding process. The scale factors may be modified such that said watermark signal is embedded in said encoded signal.

The watermark signal may be associated with a predetermined user or user device, and may therefore be used in forensic watermarking applications.

The watermark signal may be embedded by said user or at said user device. The watermark signal may alternatively be embedded prior to transmission of said information signal to said user or user device.

The method may further comprise modifying said scale factors to generate a compressed signal. At least some of said scale factors may be generated by modification of initial scale factors such that said information signal is an encrypted signal. Modifying at least one of said scale factors based upon said watermark signal may comprise modifying at least one of said scale factors to decrypt said information signal and to embed said watermark signal in said information signal. Modifying at least one of said scale factors based upon said watermark signal may comprise modifying at least one of said scale factors to decrypt said information signal. The decryption and watermark embedding may be carried out in a single operation. The decryption and watermark embedding may be carried out at the same time.

The invention further provides a computer program configured to carry out the method described above. Such a computer program can be carried on a tangible or intangible carrier medium.

BRIEF DESCRIPTION OF DRAWINGS

Embodiments of the invention will now be described, by way of example, with reference to the accompanying drawings, in which: Fig. 1 is a schematic illustration of a process for encoding an audio signal;

Fig. 2 is a schematic illustration of a process for decoding an audio signal encoded in accordance with the process of Fig. 1;

Fig. 3 is a schematic illustration of the process of Fig. 2 modified to add a watermark to the decoded signal; Fig. 4 is a schematic illustration of a process for generating a watermark signal;

Fig. 5 is a schematic illustration of a process for filtering a watermarked signal in which a watermark is embedded to allow watermark detection; Fig. 6 is a schematic illustration of a process for detecting a watermark in a signal filtered using the process of Fig. 5;

Fig. 7 is a schematic illustration of a process for encoding and encrypting an audio signal;

Fig. 8 is a schematic illustration of a process for decoding, decrypting and watermarking an audio signal encoded and encrypted using the process of Fig. 7;

Fig. 9 is a schematic illustration of a scenario in which an embodiment of the invention can be effectively employed; and

Fig. 10 is a schematic illustration of a process for modifying the bitrate of an encoded signal, and embedding a watermark in that encoded signal.

DECRIPTION OF EMBODIMENTS

Referring to Figure 1 , a process for encoding data according to the Advanced Audio Coding Standard (AAC) is shown. Such a process is often applied to digital media signals in order to encode such signals for transmission and storage. A time-domain data signal x[n] 1 is provided as input. The data signal 1 is made up of a plurality of samples. Frames can be defined by taking a number of successive samples from the time-domain data signal 1 , the number being determined by a frame size N defining a number of samples included within a frame. Frames may be defined so as to have a degree of overlap, such that at least some samples are included in more than one frame.

N In one embodiment, each frame k overlaps the previous frame by — samples.

For example, in the case where JV= 1024 and frame k=\ extends from t=l to f=1024, frame k=2 would extend from t=513 to f=1536, frame k=3 would extend from f=1025 to f=2048, frame k=4 would extend from t= 1537 to f=2560 and so on. Frames of samples taken from the data signal 1 are shaped using a Hanning Window before being processed. Suitably shaped frames of samples taken from the time-domain data signal 1 are input to a Modified Discrete Cosine Transform (MDCT) 2 which outputs transformed data X_k[m] 3. While the data signal 1 is a time-domain signal, the transformed data 3 is a frequency-domain signal. The n^th value in the k^th frame of time-domain data signal 1 is denoted x_k[n] . The result of the MDCT 2 for a particular frame of the input signal is represented by M values, such that the m^th value of the k^th frame of transformed data 3 is denoted X_k[m] .

The transformed data 3 is input to a loop comprising a quantization process 4 and a rate/distortion control process 5. The quantization process 4 quantizes the transformed data 3 to produce quantized data

6. The quantization is performed by allocating the values of a frame X_k of the transformed data into P scale factor frequency bands, such that

each scale factor frequency band comprises — values of frame X_k . Each scale factor

frequency band X_{k p} has an associated scale factor SF_k[p] which is determined by the rate/distortion control process 5.

Transformed data 3 is quantized by the quantization process 4 according to equation (1):

Q{X_k[m]) = _Sign(X_k[m]) ' - μ (i)

where

X_k[m] is the m^th value of frame k of transformed data 3; sign is a function which provides an output indicative of the sign of its input;

Q{X_k[m]) is a value representing a quantized value of X_k[m] ; μ is a constant offset value; and

SF_k[p] is the scale factor associated with the scale factor frequency band of which X_k[m] is a member where/? can be given by equation (2):

mP p = floorl (2) M

Quantized data 6 is input to the rate/distortion control process 5, where the quantized data 6 is analyzed to produce scale factors SF _k 7 such that the distortion of quantized data 6 as compared to data signal 1 is reduced. Distortion of quantized data 5 can be determined using any appropriate method.

Production of scale factors SF_k 7 and quantization of transformed data 3 into quantized data 6 is repeated until changes in SF_k 7 produce no decrease in distortion of quantized data 6.

From the preceding description it will be appreciated that the quantization process 4 and the rate/distortion control process 5 together produce quantized data 6 and scale factors 7 which appropriately represent the transformed data 3. The quantized data 6 is input to a Huffman coder 8 which appropriately encodes the quantized data 6 to produce encoded data 9. The encoded data 9 and the scale factors 7 are together input to a multiplexer 10 the output of which is a bitstream signal b_x 11 providing an encoded representation of the input signal 1.

A process for decoding the bitstream signal 11 to produce a signal approximating the input signal is now described with reference to Figure 2. The bitstream signal b_x 11 is input to a de-multiplexer 12 which outputs

Huffman encoded data 13 and scale factors 14. The Huffman encoded data 13 is input to a Huffman decoder 15 which outputs quantized data 16. The quantized data 16 is input to an inverse quantization process 17, together with the scale factors 14. The inverse quantization process 17 generates transformed data 18, which is an approximation of transformed data 3. The inverse quantization process is based upon equation (3):

X_k [m] = sign{Q{X_k [m])) ^■

μ )ϊ ^• 2 ⁴ (3)

where XAPΛ is ^a transformed data 18 value which is an approximation of X_k[m] .

The transformed data 18 is input to an inverse MDCT 19 which outputs an output signal 20.

A process for embedding a watermark in an AAC encoded signal is now described. A watermark is embedded by modifying only scale factors of the bitstream signal 11. The watermark is embedded by modifying scale factors according to equation (4): _{w r} , lSF_k[p] + w_mt[k]if p e FB

SFM = \ _{w r ! Λ} • (4)

[ SF_k[p] otherwise

where:

SFkIp] is the pth scale factor in the kth frame; Wint[k] is a watermark value for the kth frame; and

FB are one or more frequency bands having associated scale factors which are to be affected by the watermark.

That is, in each frame one or more predetermined scale factors are modified by a watermark, such that a signal having an embedded watermark is generated through modification of the scale factors.

Note that the watermark can be embedded as part of the AAC decoding process In such a case the process of Figure 2 is modified as shown in Figure 3. It can be seen that a watermark generator 21 is provided. Scale factors 14 output from the demultiplexer 12 are combined with data output from the watermark generator 21 at a combination unit 22. The modified scale factors SF' are then input to the inverse quantization process 17 which carries out processing as described above with reference to Figure 2, although here the dequantization process outputs data Y_k[m] which includes a watermark. Operation of the dequantization process where a scale factor is modified by the watermark is expressed by equation (5):

Y_k[m] = sign{Q{X_k[m])y

(5)

Substituting equation (4) into equation (5) gives:

FB

^m

A Maclaurin series expansion for the expression 2 ⁴ can be generated and higher order terms of the expansion can be eliminated to provide an approximation: 2 ⁴ ~ l+aW_mt[k] (V)

where α = — = 0.173

Substituting equation (7) into equation (6) gives:

Given equation (3):

Equation (8) can be rewritten as:

X_k[m] - (l +aW_mt[k]) if _PG FB

Y_k[m] = (9) X_k [m] otherwise

It is the result of equation (9) which is input to the inverse MDCT 19. The inverse MDCT operates as described above and in accordance with equation (10):

where

N M = — ; and

N is a window length; The smoothness of operation of the inverse MDCT 19 is improved by using a window function s which is multiplied with the output of equation (10): y [n\ = y_^n + N]s[n + N] + y_t [n]s[n],n = 0,...,N -I (11)

Having described operation of a process for embedding a watermark by modifying scale factors, generation of watermarks is now described, first with reference to Figure 4. The process of Figure 4 operates to create a watermark w_mt which is used in the process described above so as to modify the short time envelope of the signal to which the watermark is to be applied. In general terms the watermark w_mt comprises a sequence of values. Each value of the sequence of values is used to modify a scale factor associated with a particular frequency band in a particular frame. A sequence of frames is processed in such a way that successive values from the sequence of values are used to modify scale factors associated with the particular frequency band in successive frames.

Referring to Figure 4, a first pseudo random number generator (PRNG) 23 is provided with a seed 24. The first PRNG 23 outputs a sequence of values wo according to equation (12):

w_o[i]G [-l,l],i = O,l,...,L_w -l (12)

where L_w is the length of the sequence w₀.

A seed 25 is provided to a second PRNG 26 which outputs a sequence of values w_p according to equation (13):

w_p[i]G [-l,l],i = O,l,...,L_w -l (13)

The sequence w_p is circularly shifted at a shift block 27 by an amount C where:

C = GP (14)

where G is a constant; and

P is a payload of the watermark to be created. The shift block outputs a further sequence wj. The sequence W₀ is interleaved with a sequence of equal values but opposing signs. Accordingly, the sequence wo is input to a multiplier 28 where values of wo are multiplied by -1 to generate a sequence - w₀. The sequences wo and - wo are then input to a first multiplexer 29 which outputs an interleaved sequence w_mto- Similarly, the sequence W₁ is interleaved with a sequence of equal values but opposing signs. Accordingly, the sequence W₁ is input to a multiplier 30 where values of W₁ are multiplied by -1 to generate a sequence - W₁. The sequences W₁ and - W₁ are then input to a second multiplexer 31 which outputs an interleaved sequence w_mt i .

The interleaved signals w_ιnto and w_mt \ are input to a third multiplexer 32 which outputs an interleaved sequence w_mt. The sequence w_int3% is given by equation (15):

w_o[i] = k[0], W₁[O] -W₀[O] -w_λ[0l...,w₀[L_w - I]₉W₁[Z₁₁, -I] -W₀[Z₁₁, -I] -W₁[Z₁₁, -1]}

(15)

Having generated an interleaved sequence as shown in Figure 4, the resulting interleaved sequence is upsampled by a factor T_s at an upsampling unit 33 and may be windowed using an appropriate Planning window function which is multiplied by the interleaved sequence values at a convolution unit 34.

An example of the generation of a watermark signal as described with reference to Figure 4 is now described, where W₀ = [-1, I]; and wi = [1, 1].

The sequence W₁ is created by an appropriate shift of a sequence w_p in the manner described above. Interleaved sequences W^₀ and w_mt \ are then generated by appropriate multiplexing. Here: w_ιnto = [-1, 1, 1, -1]; and W^ ₁ = [I, -1, 1, -1]]

The sequences W^₀ and w_mt \ are then multiplexed to generated an interleaved signal: w_mt= [-1, 1, 1, -1, 1, 1, -1, -1]

The sequence w_mt is convolved with a Hanning window in the manner described above. The payload size that can be represented by a sequence w_mt created from sequences wo and w; in the manner described above is given by equation (16):

L β = log₂| ceiling] -g (16)

where β is the payload size;

L_w is the length of the sequences wo and W₁;

G is a constant as described above; and ceiling is a function which rounds a fraction up to the nearest integer. The payload size can be increased by generating a watermark based upon a plurality of circularly shifted sequences. In such a case, the payload size T for N_s sequences becomes:

r = (JV, -l)β (17)

Having created an interleaved sequence in the manner described above, this can be used to modify scale factors as described with reference to the process of Figure 3. The interleaved sequence is used to modify scale factors associated with a sequence of frames, such that each value of the interleaved sequence is used to modify a scale factor of a respective frame.

Having embedded a watermark in a signal by appropriate modification of scale factors, it will be appreciated that a method is required to extract that watermark for verification purposes, and such a method is now described with reference to Figures 5 and 6. A signal Y' in which a watermark is embedded is input to an envelope discrimination circuit 39. The signal Y' is input to a bandpass filter 40 which outputs a filtered signal Y'b41 before being multiplied by itself (i.e. squared) by a multiplier 42. The multiplier outputs a squared signal 43 which is input to a lowpass filter 44 which outputs a lowpass filtered signal 45, which is downsampled by a downsampler 46 to produce a signal w'_int47 which is an estimate of the watermark. Referring to Figure 6, the signal w'_int47 is input to a demultiplexer 48 which outputs two signals w'_mto 49 and w'_int i 50. The two signals w'_mto 49 and w'_int i 50 are input to respective demultiplexers 51, 52. The demultiplexer 51 outputs signals w'_o 53a and -w'_o 53b. The demultiplexer 53 outputs signals w'i 54a and -w'i 54b. The outputs of the demultiplexers 51, 52 are respectively input to respective adders 55, 56. The outputs of the adders are input to respective correlation blocks 55 a, 56a which determine whether the output signals could be derived by shifting the sequences used for watermark creation. It was indicated above that a greater payload could be included within a watermark by using a greater number of sequences for watermark creation. It should however be noted that increasing the number of sequences also increases the length of audio clip required to detect the presence or absence of a watermark. Specifically, the required length of clip is given by:

L_dψ = 2N_sL_wM (18)

Where L_w is 512 and β is 7, a 47.5 second clip is required for watermark detection at a sampling frequency of 44100 Hz. If a 35 bit payload is used (i.e. if β is 35), which is common in forensic watermarking applications, the detection period increases to 142.5 seconds.

In order to increase the possible payload without requiring a longer sample for proper watermark detection, an alternative method may be used. Specifically different watermark sequences are associated with different frequency bands, and accordingly different scale factors are modified using different watermark sequences. For example three different frequency bands may be used, and their scale factors can be modified according to equation (19):

where: FBQ,

are respective frequency bands; and w_mt_₀ , w_mt__j , w_mt_₂ are watermark sequences used to modify scale factors associated with the respective frequency bands. Each of the watermark sequences of equation (19) is created by interleaving a pair of sequences. The sequences to be interleaved are themselves created by interleaving a sequence with a sequence of equal values but opposing signs.

Using the method described with reference to equation (19) six initial sequences can be used to create the three watermark sequences which are used to modify scale factors associated with particular frequency bands. In this way, six sequences can be used to allow a 35 -bit payload to be detected from a clip of 47.5 seconds duration, where other parameters are as described above.

Watermark strength can be modified to ensure that a signal including a watermark does not have unacceptably undesirable artefacts. It has been found that allowing watermark values to take values of -1 or 1 provides acceptable performance, that is, added artefacts are not unacceptable. However, in particularly sensitive applications, values of-1 and 0 or 0 and 1 may be used.

In many applications, frames of AAC data comprise 2048 samples. However AAC signals sometimes include short windows including only 256 samples. Short windows always occur in multiples of eight. In order to allow windows of both 2048 samples and 256 samples to be processed analogously, it is sometimes desirable to embed a common watermark in all eight consecutive short windows. Alternatively, given that short windows occur rarely, no modification of such windows may be carried out without any detriment to overall performance.

The preceding description has been concerned with embedding of watermarks in AAC encoded data. It is to be noted that embodiments of the invention are not limited to AAC encoded data, but are instead widely applicable to any encoding scheme employing scale factors. In particular the methods described can be applied to data encoded using the MP3 and AC3 standards.

The preceding description has been concerned with embedding a watermark during a decoding process. It will be appreciated that the invention is not limited to the embedding of watermarks in this way. Specifically, in some embodiments the watermark is embedded in encoded data using the process described above. That is, an encoded signal is processed so as to produce an encoded signal in which a watermark is embedded.

Alternatively, a watermark may be embedded in an information signal as part of the encoding process. That is, a non-encoded signal may be processed to generate a partially-encoded signal, and that partially encoded signal may be processed so as to cause said watermark to be embedded. Various ways in which the method described above can be applied are now described in further detail.

In a process shown in Figure 7, the encoding process of Figure 1 is modified such that scale factors are generated to include a masker sequence r[k]. The masker sequence r[k] comprises a plurality of values and respective values of the masker sequence r[k] are used to modify a scale factor associated with a particular frequency band in each of a plurality of frames of data. Values for the masker sequence are selected such that the original signal cannot be properly accessed without removal of the masker sequence, i.e. it is encrypted. For example, where the original signal is a sound file, after addition of the masker sequence unacceptable artefacts are present when the file is played back.

Referring to Figure 7, it can be seen that scale factors are input to an adder 60 into which the masker sequence r[k] is also input. Modified scale factors SF "^ 61 are output from the adder 60 and then passed to the multiplexer 10 which now outputs an encrypted signal E{b_x} having modified scale factors. Where scale factors are modified using the encoding process of Figure 7 it will be appreciated that such modification must be reversed during a decoding process. Referring to Figure 8, it can be seen that the process shown in Figure 3 has been modified to take into account the encryption of Figure 7. Specifically, a signal r_w [k] is input to an adder 22. Obviously, if the signal r_w [k] were the inverse of the masker sequence r[k] of Figure 7, it would cause the encrypted encoded signal to be decrypted during decoding. However, in accordance with an aspect of the invention, the signal r_w [k] is defined by equation (20):

r_w[k] = w_mt[k] - r[k] (20)

That is, r_w[k] is configured such that its addition to a sequence of scale factors which have been modified using the masker sequence r[k] reverses the modifications caused by the masker sequence while at the same time embedding the watermark w_mt[£] .

Figure 9 shows a scenario in which encoding according to the process of Figure 7 and decoding according to the process of Figure 8 can be effectively used. A content owner 65 may wish to make content available to a plurality of users. However the content owner 65 wishes to ensure that content is accessible only to authorized users, and further wishes to ensure that content provided to a particular authorized user can be traced as having been provided to that user. When content is provided to a particular authorized user i 66, it is encrypted by addition of a masker sequence r_x so as to generate a signal E{b_x}j. The masker sequence r_x is provided to an independent service provider 67, and processed so as to generate a sequence r_m which combines the masker sequence r_λ with a watermark sequence W₁ which is generated as described above with reference to Figure 4. The authorized user 66 is then able to both decrypt the received signal and embed a watermark within the signal as described above with reference to equation (20).

It will be appreciated that in alternative embodiments of the invention a watermark specific to a particular authorized user may be embedded by the content owner 65 prior to transmission of the content to the authorized user 66. In such a case, the authorized user 66 is simply provided with details of the masker sequence used for encryption so as to allow decryption. Such a masker sequence can be provided either by the content owner 65 or by the service provider 67. Although a watermark can be embedded by the content owner 65 in this way, it is often preferred that the watermark is embedded as the content by individual authorized users, so as to distribute the necessary processing.

In some embodiments an encoded signal is processed so as to provide an encoded signal in which a watermark is embedded. The watermark is embedded using the process described above. Embedding of the watermark may be carried out as part of a process which further modifies the scale factors so as to modify the bit rate of the encoded signal. Embedding of a watermark together with bit rate modification is now described with reference to Figure 10.

A bit stream signal b_x 70 is input to a demultimp lexer 71 which produces as output quantized data Q{x) 72 and scale factors SF 73. The scale factors are input to a modification block 74 which outputs modified scale factors SF' 75. The modification block 74 receives data from a watermark generation block 76, together with data A_{1 2}77 indicating a desired change in bit rate. In such a case the modified scale factors SF' 75 are given by equation (21):

SF'= Φ((SF + w_mt),A_U2) (21)

where Φ is a function which modifies scale factors so as to cause a modification of bit rate.

The quantized data Q{x) 72 is input to a requantize block 77 which requantizes the data to generated requantized data Q\x) 78 in accordance with equation (22):

where Ψ is a requantization function. The requantized data 78 and the modified scale factors 75 are input to a multiplexer 79 which generates a modified bitstream signal b_x ' 80. The modified bitstream signal 80 has an embedded watermark, and a bitrate determined by the data A_{1 2}Il.

Embodiments of the invention have been described in some detail above. It will however be appreciated that various modifications can be made to the described embodiments without departing from the scope of the appended claims.

In summary, a method for generating a watermark signal and embedding the watermark signal in an information signal is disclosed. The information signal comprises a plurality of scale factors (SFt) each of which is associated with a range of frequencies. The method comprises generating (21) said watermark signal by combining first and second sequences of values, at least one of said first and second sequences of values being generated by shifting an initial sequence of values in a manner determined by a payload. At least one of said scale factors (SFt) is modified (22) based upon said watermark signal.

In an application of the invention, a bitstream is encrypted by modifying the scale factors at a server side. At the client side, the bitstream is decrypted by modifying the scale factors such that a watermark is left in the decrypted signal.

Claims

CLAIMS:

1. A method for generating a watermark signal and embedding the watermark signal in an information signal, the information signal comprising a plurality of scale factors each of which is associated with a range of frequencies, the method comprising: generating said watermark signal by combining first and second sequences of values, at least one of said first and second sequences of values being generated by shifting an initial sequence of values in a manner determined by a payload; modifying at least one of said scale factors based upon said watermark signal.

2. A method according to claim 1, wherein said information signal comprises a plurality of frames, each of said frames comprises said plurality of scale factors and scale factors in each of said frames are modified based upon a respective part of said watermark signal.

3. A method according to claim 1 or 2, further comprising: generating a plurality of watermark signals by combining respective first and second sequences of values, at least one of said respective first and second sequences of values being generated by shifting an initial sequence of values in a manner determined by a payload.

4. A method according to claim 3, further comprising: modifying a plurality of said scale factors, each of said scale factors being modified based upon a predetermined watermark signal.

5. A method according to any preceding claim, wherein said information signal is an encoded signal, and said scale factors are modified when said encoded signal is decoded.

6. A method according to any one of claims 1 to 4, wherein said information signal is an encoded signal, and said scale factors are modified such that said watermark signal is embedded in said encoded signal.

7. A method according to any one of claims 1 to 4, further comprising: processing said information signal so as to generate an encoded signal; wherein said scale factors are modified during said processing.

8. A method according to any preceding claim, wherein said watermark signal is associated with a predetermined user or user device and embedded by said user or at said user device.

9. A method according to claim 8, wherein said watermark signal is embedded prior to transmission of said information signal to said user or user device.

10. A method according to any preceding claim, further comprising modifying said scale factors to modify said bit rate of said information signal.

11. A method according to any preceding claim, wherein at least some of said scale factors are generated by modification of initial scale factors such that said information signal is an encrypted signal.

12. A method according to claim 11, wherein modifying at least one of said scale factors based upon said watermark signal comprises modifying at least one of said scale factors to decrypt said information signal and to embed said watermark signal in said information signal.

13. A computer program configured to cause a processor to carry out a method according to any preceding claim.

14. An apparatus for generating a watermark and embedding the watermark in an information signal, the apparatus comprising: a memory storing processor readable instructions; and a processor configured to read and execute instructions stored in said program memory; wherein said processor readable instructions comprise instructions controlling the processor to carry out a method according to any one of claims 1 to 14.

15. Apparatus for generating a watermark signal and embedding the watermark signal in an information signal, the information signal comprising a plurality of scale factors each of which is associated with a range of frequencies, the apparatus comprising: a watermark generator configured to generate a watermark signal by combining first and second sequences of values, at least one of said first and second sequences of values being generated by shifting an initial sequence of values in a manner determined by a payload; a watermark embedder configured to modify at least one of said scale factors based upon said watermark signal.