Digital watermarks adapted to compensate for time scaling, pitch shifting and mixing
Pre-processing modules are configured to compensate for time and pitch scaling and shifting and provide compensated audio frames to a watermark detector. Audio frames are adjusted for time stretching and shrinking and for pitch shifting. Detection metrics are evaluated to identify candidates to a watermark detector. Various schemes are also detailed for tracking modifications made to audio stems mixed into audio tracks, and for accessing a history of modifications for facilitating identification of audio stems and audio tracks comprised of stems. Various approaches address interference from audio overlays added to channels of audio after embedding. One approach applies informed embedding based on phase differences between corresponding components of the channels. A detector extracts the watermark payload effectively from either additive or subtractive combination of the channels because the informed embedding ensures that the watermark survives both types of processing. Other approaches applies different polarity patterns, watermark mappings, or protocol keys to the channels. These techniques enable the watermark to survive ambient mixing, conversion to mono, as well as channel differencing to reduce interference from voice-overs and other audio overlays.
1. A method for compensating for time or pitch scaling for audio watermark detection, the method comprising:
receiving an audio watermarked signal;
for each of plural streams of the audio watermarked signal, performing a candidate time adjustment to an input frame of watermarked audio, and a candidate pitch shift adjustment to the input frame to produce a compensated audio frame;
for the compensated audio frame, measuring a detection metric; and
based on the detection metric, selecting compensated audio frames to provide for watermark detection.
2. The method of claim 1 wherein the candidate time adjustment is performed by zero padding the input frame.
3. The method of claim 2 wherein the candidate pitch shift adjustment is performed by interpolating frequency components of the input frame.
4. The method of claim 1 wherein the detection metric is a repetitive structure metric based on repetition of a generator polynomial of an error correction encoder used to encode a watermark signal in a tile mapped to embedding locations in an audio frame.
5. The method of claim 1 wherein the detection metric is a repetitive structure metric based on repetition of a watermark element state encoded in a watermark signal in a tile, and repeated in tiles mapped to embedding locations in audio frames.
6. An audio watermark detector configured to compensate for time or pitch scaling, the detector comprising:
memory for storing an audio watermarked signal;
means for generating a candidate time adjustment to an input frame of watermarked audio, and a candidate pitch shift adjustment to the input frame to produce a compensated audio frame, for each of plural streams of the audio watermarked signal;
means for computing a detection metric for the compensated audio frame; and
means for selecting compensated audio frames to provide for watermark detection based on the detection metric.
7. The detector of claim 6 wherein the candidate time adjustment is performed by zero padding the input frame.
8. The detector of claim 7 wherein the candidate pitch shift adjustment is performed by interpolating frequency components of the input frame.
9. The detector of claim 6 wherein the detection metric is a repetitive structure metric based on repetition of a generator polynomial of an error correction encoder used to encode a watermark signal in a tile mapped to embedding locations in an audio frame.
10. The detector of claim 6 wherein the detection metric is a repetitive structure metric based on repetition of a watermark element state encoded in a watermark signal in a tile, and repeated in tiles mapped to embedding locations in audio frames.
11. The audio watermark detector of claim 6 , further comprising:
a pre-processor configured to receive first and second channels of audio and compute a difference signal; and
a detector configured to receive the difference signal, and configured to receive an additive combination of the first and second channels, the detector operable to extract a watermark signal that has been encoded into both the first and second channels from the difference signal and the additive combination of the first and second channels.
12. The detector of claim 11 wherein the watermark signal is embedded in the first and second audio channels by:
evaluating phase differences between corresponding components of first and second audio channels;
based on the evaluating, adapting gain applied to a watermark applied to at least one of the corresponding components; and
inserting the watermark in the first and second audio channels, wherein the watermark signal is retained in both a conversion of the first and second channels to a mono signal by an additive combination or a subtractive combination.
13. The detector of claim 11 , wherein the watermark signal has been encoded in the first and second channels by evaluating phase differences between corresponding components of first and second audio channels, adapting gain applied to a watermark applied to at least one of the corresponding components based on the evaluating, and inserting the watermark in the first and second audio channels, wherein the watermark signal is retained in both a conversion of the first and second channels to a mono signal by an additive combination or a subtractive combination.
14. The detector of claim 11 wherein the watermark signal has been encoded in the first and second channels by encoding a watermark tile transformed by different protocol keys in the first and second channels.
15. The detector of claim 11 wherein the watermark signal has been encoded in the first and second channels by encoding a watermark tile transformed by a different polarity pattern in the first and second channels, the polarity pattern in the first channel being offset relative to the polarity pattern in the second channel.
16. The detector of claim 11 wherein the watermark signal has been encoded in the first and second channels by encoding a watermark tile transformed by a different embedding location mapping in the first and second channels.
17. The detector of claim 16 wherein the mapping comprises a different watermark resolution for the watermark tile in the first and second channels.
18. A non-transitory computer readable medium on which is stored instructions, the instructions configured to execute a method for compensating for time or pitch scaling for audio watermark detection, the method comprising:
receiving an audio watermarked signal;
for each of plural streams of the audio watermarked signal, performing a candidate time adjustment to an input frame of watermarked audio, and a candidate pitch shift adjustment to the input frame to produce a compensated audio frame;
for the compensated audio frame, measuring a detection metric; and
based on the detection metric, selecting compensated audio frames to provide for watermark detection.
19. The non-transitory computer readable medium of claim 18 comprising instructions configured to perform the candidate time adjustment by zero padding the input frame.
20. The non-transitory computer readable medium of claim 19 wherein the candidate pitch shift adjustment is performed by interpolating frequency components of the input frame.