Audio watermarking with dual watermarks

• US 7,206,649 B2
• Filed: 10/21/2004
• Issued: 04/17/2007
• Est. Priority Date: 07/15/2003
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Kirovski, Darko, Malvar, Henrique, Jakubowski, Mariusz H.
• Primary Examiner(s): Tran, Sinh
• Assistant Examiner(s): Sellers, Daniel R.
• Application Number:
  US10/970,499
• Publication Number:
  US 20050055214A1
• Time in Patent Office: 908 Days
• Field of Search: 700/94, 713/176, 726/26, 726/32, 380/200, 380/201
• US Class Current: 700/94
• CPC Class Codes: G10L 19/018 Audio watermarking, i.e. embedding inaudible data in the audio signal

Abstract

A watermark encoding system encodes an audio signal with both a strong and a weak watermark. The strong watermark identifies the content producer and is designed to survive all typical kinds of processing and malicious attacks. The weak watermark identifies the content as an original and is designed to be significantly removed as a result of most normal signal processing (other than A/D and D/A). The watermark encoding system has a converter to convert an audio signal into frequency and phase components and a mask processor to determine a hearing threshold for corresponding frequency components. The watermark encoding system also has a pattern generator to generate both the strong and weak watermarks and a watermark insertion unit to selectively insert either the strong or weak watermark into the audio signal. The watermark insertion unit adds the strong watermark to the audio signal when the signal exceeds the hearing threshold by a buffer value (e.g., 1–8 dB) and adds the weak watermark insertion unit when the signal falls below the hearing threshold by the buffer value. When the signal falls within the buffer area about the hearing threshold, the insertion unit takes no action. A watermark detecting system is equipped with a watermark detector that determines which block interval of the watermarked audio signal contains a watermark pattern and if the strong or weak watermark is present in that block interval of the signal.

Claims

1. An audio watermarking system comprising:

a pattern generator means for generating both a strong watermark and a weak watermark; and

a watermark insertion means for selectively inserting either the strong watermark or the weak watermark into one or multiple segments of an audio signal according to an audible measure of a segment having a watermark inserted therein.

2. An audio watermarking system comprising:

a processor means for determining a hearing threshold for an audio signal;

a pattern generation means for generating both a strong watermark and a weak watermark;

a watermark insertion means for inserting the strong watermark into the audio signal when the audio signal exceeds the hearing threshold and for inserting the weak watermark into the audio signal when the signal falls below the hearing threshold.

3. An audio watermark encoding system comprising:

a conversion means for converting an audio signal into magnitude and phase components;

a mask processor means for determining a hearing threshold for corresponding magnitude components;

a pattern generator means for generating both a strong watermark and a weak watermark; and

a watermark insertion means for selectively inserting one of either the strong watermark or the weak watermark into the audio signal based on whether the magnitude components exceed or fall below the hearing threshold.

4. An audio watermark encoding system as recited in claim 3, wherein the watermark insertion means is also for inserting the strong watermark when the magnitude component exceeds the hearing threshold and for inserting the weak watermark when the magnitude component falls below the hearing threshold.

5. An audio watermark encoding system as recited in claim 3, wherein the watermark insertion means is also for inserting the strong watermark when the magnitude component exceeds the hearing threshold by a predetermined amount and for inserting the weak watermark when the magnitude component falls below the hearing threshold by the predetermined amount.

6. An audio watermark encoding system as recited in claim 3, wherein the watermark insertion means is for foregoing insertion of the strong watermark or the weak watermark when the magnitude component lies within the predetermined amount above and below the hearing threshold.

7. An audio encoding system comprising:

an audio watermark encoding means as recited in claim 3; and

a compression means for compressing, wherein the compression means and the audio watermark encoding means both utilize the magnitude components.

8. An operating system comprising:

a conversion means for converting an audio signal into magnitude and phase components;

a mask processor means for determining a hearing threshold for corresponding magnitude components;

a pattern generator means for generating both a strong watermark and a weak watermark; and

a watermark insertion means for selectively inserting one of either the strong watermark or the weak watermark into the audio signal based on whether the magnitude components exceed or fall below the hearing threshold.

9. A watermark insertion unit, comprising:

an input receiving means for receiving frequency magnitude components of an audio signal, hearing thresholds derived from the magnitude components, strong watermark values, and weak watermark values; and

multiple insertion operation means for selectively combining the magnitude components and one of either the strong watermark values or the weak watermark values depending upon whether the magnitude components exceed or fall below the hearing thresholds.

10. An audio watermark detection system, comprising:

an input module means for receiving a watermarked audio signal;

a synchronization module means for determining which portion of the watermarked audio signal might contain a watermark; and

a correlation module means for detecting whether a watermark is present in the portion of the watermarked audio signal that the synchronization module means determines might contain a watermark and, if a watermark is detected, the correlation module means is also for detecting whether that watermark is either a strong watermark or a weak watermark.

11. An audio watermark detection system as recited in claim 10 further comprising a computation means for computing a correlation value from the watermarked audio signal and the strong watermark that tends toward a first value when the strong watermark is present and a second value when the strong watermark is not present.

12. An audio watermark detection system as recited in claim 10 further comprising a computation means for computing a correlation value from the watermarked audio signal and the weak watermark that tends toward a first value when the weak watermark is present and a second value when the weak watermark is not present.

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

a computation means for computing a correlation value from the watermarked audio signal and one of either the strong watermark or the weak watermark;

a determination means for determining that said one strong watermark or weak watermark is present when the correlation value exceeds a predetermined threshold plus a random amount.

14. An operating system comprising:

an input module means for receiving a watermarked audio signal;

a synchronization module means for determining which portion of the watermarked audio signal might contain a watermark;

a correlation module means for detecting whether a watermark is present in the portion of the watermarked audio signal that the synchronization module means determines might contain a watermark;

an identification means for identifying a detected watermark as either a strong watermark or a weak watermark.

15. An audio watermark detection system comprising:

a pattern generation means for generating both a strong watermark and a weak watermark;

a watermark detection means for detecting whether a watermark is present in a portion of the watermarked audio signal, wherein the detecting is based upon computing correlation values from the watermarked audio signal and each of the strong watermark and the weak watermark;

an identification means for identifying a detected watermark as either a strong watermark or a weak watermark, wherein the identifying is based upon whether the correlation values exceed a predetermined threshold.

16. An audio watermark detection system comprising:

a random operation means for generating a random value;

a pattern generation means for generating both a strong watermark and a weak watermark;

a watermark detection means for detecting whether a watermark is present in a portion of the watermarked audio signal;

a computing means for computing correlation values from the watermarked audio signal and each of the strong watermark and the weak watermark;

an identification means for identifying a detected watermark as either a strong watermark or a weak watermark, the identifying being based upon whether the correlation values exceed a predetermined threshold plus the random value.

17. One or more computer-readable media having computer-executable instructions that, when executed by a computer, performs steps comprising:

a comparing step for comparing samples of an audio signal to a hearing threshold;

a first watermarking step for watermarking samples exceeding the hearing threshold with a strong watermark; and

a second watermarking step for watermarking samples falling below the hearing threshold with a weak watermark.

18. One or more computer media as recited in claim 17, wherein the first and second watermarking steps comprise:

a first watermarking step portion for watermarking samples exceeding the hearing threshold plus a buffer value with a strong watermark;

a second watermarking step portion for watermarking samples falling below the hearing threshold by less than the buffer value a with a weak watermark; and

a third watermarking step portion for leaving samples lying within the buffer value above and below the hearing threshold without a watermark.

19. One or more computer media as recited in claim 17, wherein the steps further comprise a detecting step for detecting the strong watermark and the weak watermark in the audio signal.

20. One or more computer media as recited in claim 19, wherein the detecting step for detecting comprises a computing step for computing a correlation value from the audio signal and the strong watermark, the correlation value tending toward a first value when the strong watermark is present and a second value when the strong watermark is not present.

21. One or more computer media as recited in claim 19, wherein the detecting step for detecting comprises a computing step for computing a correlation value from the audio signal and the weak watermark, the correlation value tending toward a first value when the weak watermark is present and a second value when the weak watermark is not present.

22. One or more computer media as recited in claim 17, wherein the steps further comprise:

a computing step for computing a correlation value from the audio signal and one of the strong watermark or the weak watermark; and

a determining step for determining that said one strong watermark or weak watermark is present when the correlation value exceeds a predetermined threshold plus a random amount.

23. One or more computer media as recited in claim 17, wherein the steps further comprise:

a computing step for computing a correlation value from the audio signal and one of either the strong watermark or the weak watermark; and

a determining step for determining that either said one strong watermark or said one weak watermark is present when the correlation value exceeds a predetermined threshold plus a random amount.

24. An audio watermarking system comprising:

a pattern generation means for generating both a strong watermark and a weak watermark; and

a watermark insertion means for inserting the strong watermark into one or more first segments of an audio signal and for inserting the weak watermark into one or more second segments of the audio signal, wherein the first and second segments are separate from each other, wherein the watermark insertion means is also for selectively choosing segments for insertion of the weak watermark according to an audible measure of the segments.