Watermark detection via cardinality-scaled correlation

• US 6,738,744 B2
• Filed: 12/08/2000
• Issued: 05/18/2004
• Est. Priority Date: 12/08/2000
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Kirovski, Darko, Malvar, Henrique
• Primary Examiner(s): ABEBE, DANIEL DEMELASH
• Application Number:
  US09/733,576
• Publication Number:
  US 20020107691A1
• Time in Patent Office: 1,257 Days
• Field of Search: 704/273, 713/193
• US Class Current: 704/273
• CPC Class Codes: G10L 19/018 Audio watermarking, i.e. embedding inaudible data in the audio signal

Abstract

Described herein is an audio watermarking technology for detecting watermarks in audio signals, such as a music clip. The watermark identifies the content producer, providing a signature that is embedded in the audio signal and cannot be removed. The watermark is designed to survive all typical kinds of processing and all types of malicious attacks that attempt to remove or modify the watermark from the signal. The implementations of the watermark detecting system, described herein, support quick, efficient, and accurate detection of watermarks by the specifically designed watermark detecting system. In one described implementation, a watermark detecting system employs an improved normalized covariance test to determine the presence of a watermark using less expensive materials (hardware), quicker calculations, and a more accurate test (than the original correlation test). In other described implementations, a watermark detecting system employs a cepstrum filter and dynamic processing to minimize the affect of the “noise” in the watermarked signal. The “noise” is the original content of the signal before such signal was watermarked. In still another described implementation, a watermark detecting system employs a mechanism for random detection threshold so that the act of watermark detection does not provide decipherable clues to a digital pirate as to the value or location of the embedded watermark.

Claims

1. An audio watermark detection system, comprising:

a pattern generator to generate a watermark (w) comprised of two defined values (a and b); and

a correlation module to detect whether the watermark is present in a watermarked audio signal (y), wherein the correlation module computes a cardinality-scaled correlation (CSC) value from the watermarked audio signal and from the watermark based upon;

a difference between the sum of signal samples of y for which the corresponding watermark bit w matches a divided by the cardinality of watermark samples matching a, and the sum of signal samples of y for which the corresponding watermark bit w matches b divided by the cardinality watermark samples equal to b.

2. A system as recited in claim 1, wherein a is one (1) and b is zero (0).

3. A system as recited in claim 1, wherein the watermarked audio signal has a high ratio of noise to the watermark, the system further comprising a watermark pre-processor to reduce such noise in the watermarked signal.

4. A system as recited in claim 3, wherein the pre-processor cepstrum filters the watermarked signal.

5. A system as recited in claim 3, wherein the pre-processor non-linearly modifies the watermarked signal such that the low-energy frequency amplitudes are attenuated and the high-energy frequency amplitudes are amplified.

6. A system as recited in claim 3, wherein the CSC value computed by the correlation module tends toward a first value when the watermark is present and towards a second value when the watermark is not present.

7. A system as recited in claim 6, wherein the first value is one (1) and the second value is zero (0).

8. A system as recited in claim 1, further comprising:

a random operator for generating a random value; and

the correlation module computes the CSC value from the watermarked audio signal and detects the presence of the watermark based on whether the CSC value exceed a predetermined threshold plus the random value.

9. An operating system comprising an audio watermark detection system as recited in claim 1.

10. An audio watermark detection system comprising:

a pattern generator to generate a watermark encoded as a sequence of values selected from a set of values; and

a watermark detector to detect presence of the watermark encoded into the frequency domain of an digital signal, wherein the detector detects the presence of the watermark by tracking;

sum of occurrences of given values in the signal conditioned upon the watermark and the signal; and

cardinality of such occurrences of the same given values in the watermark itself.

11. An audio watermark detection system as recited in claim 10, wherein the watermark detector computes a cardinality-scaled correlation (CSC) value from the digital signal and of the watermark and detects the presence of the watermark based on whether the CSC value exceeds a predetermined threshold.

12. An audio watermark detection system as recited in claim 10, further comprising:

a random operator for generating a random value; and

the watermark detector computes cardinality-scaled correlation (CSC) values from the digital signal and each of the watermark and detects the presence of the watermark based on whether the CSC value exceed a predetermined threshold plus the random value.

13. A method of detecting presence of a watermark in an audio signal, the method comprising:

generating a watermark a watermark (w) comprised of two defined values (a and b); and

computing a cardinality-scaled correlation (CSC) value to detect whether the watermark is present in a watermarked audio signal (y), wherein the CSC value is computed from the watermarked audio signal and from the watermark based upon;

$\frac{\mathrm{sum}<br><br>\left(y|w=a\right)}{\mathrm{card}<br><br>\left(w=a\right)}-\frac{\mathrm{sum}<br><br>\left(y|w=b\right)}{\mathrm{card}<br><br>\left(w=b\right)}$

14. A method as recited in claim 13, wherein a is one (1) and b is zero (0).

15. A method as recited in claim 13, wherein the watermarked audio signal has a high ratio of noise to the watermark, the method further comprising noise-reduction pre-processing of the watermarked signal to reduce such noise.

16. A method as recited in claim 15, wherein the pre-processing includes cepstrum filtering of the watermarked signal.

17. A method as recited in claim 15, wherein the pre-processing includes non-linearly modifying the watermarked signal such that the low-energy frequency amplitudes are attenuated and the high-energy frequency amplitudes are amplified.

18. A method as recited in claim 13, further comprising detecting presence of watermark based upon whether the CSC value exceed a predetermined threshold.

19. A method as recited in claim 13, further comprising detecting presence of the watermark by examining the CSC value computed by the computing, such that the CSC value tends toward a first value when the watermark is present and towards a second value when the watermark is not present.

20. A method as recited in claim 19, wherein the first value is one (1) and the second value is zero (0).

21. A method as recited in claim 13, further comprising:

generating a random value; and

detecting the presence of the watermark based upon whether the CSC value exceed a predetermined threshold plus the random value.

22. A computer-readable medium having computer-executable instructions that, when executed by a computer, performs the method as recited in claim 13.

23. A computer-readable medium having computer-executable instructions that, when executed by a computer, perform a method of detecting a watermark in an audio signal, the method comprising:

generating a watermark encoded as a sequence of values selected from a set of values; and

detecting presence of the watermark encoded into the frequency domain of the digital signal, wherein the presence of the watermark is determined by tracking;

sum of occurrences of given values in the signal conditioned upon the watermark and the signal; and

cardinality of such occurrences of the same given values in the watermark itself

to calculate a cardinality-scaled correlation (CSC) value which indicates the presence of the watermark if the CSC value exceeds a threshold.

24. A modulated signal indicating whether a watermark is present within an audio signal, the modulated signal generated in accordance with the following acts:

generating a watermark; and

detecting presence of the watermark encoded into the frequency domain of the digital signal, wherein the presence of the watermark is determined by tracking;

sum of occurrences of given values in the signal conditioned upon the watermark and the signal; and

cardinality of such occurrences of the same given values in the watermark itself;

to calculate a cardinality-scaled correlation (CSC) value which indicates the presence of the watermark if the CSC value exceeds a threshold.

25. A watermark detection system comprising:

a pattern generator to generate a watermark;

a random operator for generating a random value; and

a correlation module to detect whether the watermark is present in an audio signal, wherein the correlation module;

computes a cardinality-scaled correlation (CSC) value from the audio signal and from the watermark; and

detects the presence of the watermark based on whether the CSC value exceed a predetermined threshold plus the random value.

26. A watermark detection method comprising:

generating a watermark;

generating a random value; and

determining whether the watermark is present in an audio signal by computing a cardinality-scaled correlation (CSC) value from the audio signal and from the watermark and detecting the presence of the watermark based on whether the CSC value exceed a predetermined threshold plus the random value.

27. A computer-readable medium having computer-executable instructions that, when executed by a computer, performs the method as recited in claim 26.

28. An audio watermark detection system, comprising:

a pattern generator to generate a watermark (w) comprised of two defined values (a and b); and

a correlation module to detect whether the watermark is present in a watermarked audio signal (y), wherein the correlation module computes a cardinality-scaled correlation (CSC) value from the watermarked audio signal and from the watermark based upon;

$\frac{\mathrm{sum}<br><br>\left(y|w=a\right)}{\mathrm{card}<br><br>\left(w=a\right)}-\frac{\mathrm{sum}<br><br>\left(y|w=b\right)}{\mathrm{card}<br><br>\left(w=b\right)}$

29. An audio watermark detection system, comprising:

a pattern generator to generate a watermark (w) comprised of two defined values (a and b); and

a correlation module to detect whether the watermark is present in a watermarked audio signal (y), by computing a cardinality-scaled correlation (CSC) value from the watermarked audio signal and from the watermark, wherein the CSC value is computed as a difference between a first CSC value based on an assumption that w=a and a second CSC value based on an assumption that w=b.

30. A system as recited in claim 29, wherein a is one (1) and b is zero (0).

31. A system as recited in claim 29, wherein the watermarked audio signal has a high ratio of noise to the watermark, the system further comprising a watermark pre-processor to reduce such noise in the watermarked signal.

32. A system as recited in claim 31, wherein the pre-processor cepstrum filters the watermarked signal.

33. A system as recited in claim 31, wherein the pre-processor non-linearly modifies the watermarked signal such that the low-energy frequency amplitudes are attenuated and the high-energy frequency amplitudes are amplified.

34. A system as recited in claim 31, wherein the CSC value computed by the correlation module tends toward a first value when the watermark is present and towards a second value when the watermark is not present.

35. A system as recited in claim 34, wherein the first value is one (1) and the second value is zero (0).

36. A system as recited in claim 29, further comprising:

a random operator for generating a random value; and

the correlation module computes the CSC value from the watermarked audio signal and detects the presence of the watermark based on whether the CSC value exceed a predetermined threshold plus the random value.

37. An operating system comprising an audio watermark detection system as recited in claim 29.