Methods, Systems, and Sub-Combinations Useful in Media Identification

• US 20070174059A1
• Filed: 01/02/2007
• Published: 07/26/2007
• Est. Priority Date: 05/16/1996
• Status: Active Grant

• Current Assignee: Digimarc Corporation
• Original Assignee: Digimarc Corporation
• Inventors: Rhoads, Geoffrey, Brunk, Hugh, Levy, Kenneth
• Granted Patent: US 7,930,546 B2
• Time in Patent Office: Days
• US Class Current: 704/273
• CPC Class Codes: G10L 19/018 Audio watermarking, i.e. embedding inaudible data in the audio signal
• CPC Class Codes: G10L 25/48 specially adapted for particular use

Abstract

Identification of media content, such as audio, can be performed through use of watermarking or fingerprinting (aka content signature) technologies. Aspects of these technologies may be combined to advantageous effect. For example, in dealing with the problem of fingerprint errors arising from object distortion, operations known from digital watermarking systems can be employed.

Claims

1. A method useful in identifying media content, the content including audio data, the method comprising:

pre-processing the audio data;

deriving characteristic audio fingerprint data from the pre-processed data; and

sending the characteristic audio fingerprint data to a database, which can identify earlier-stored data that corresponds to said derived audio fingerprint data to thereby identify content that matches said media content;

wherein said pre-processing comprises performing a log-mapping process based on a frequency-domain representation of said audio data to yield logarithmically-sampled data, prior to deriving said characteristic audio fingerprint data.

2. The method of claim 1, wherein said pre-processing further includes transforming the audio data to a frequency domain representation.

3. The method of claim 1, wherein said pre-processing further includes generating audio power spectrum data corresponding to said frequency domain representation, and performing a log-mapping process based on said audio power spectrum data.

4. The method of claim 1, wherein said pre-processing comprises identifying segments of the audio data, and the deriving comprises deriving characteristic audio fingerprint data for each of said segments.

5. The method of claim 1 that further includes determining, from said logarithmically-sampled data, a distortion to which said audio data has been subjected.

6. The method of claim 5 that includes compensating said audio data for said determined distortion prior to deriving said characteristic fingerprint data.

7. The method of claim 5 that includes taking said distortion into account when deriving said characteristic fingerprint data.

8. The method of claim 1 wherein said logarithmically sampled data includes a hash.

9. The method of claim 1 including correlating the logarithmically sampled data with one or more pre-existing signals.

10. A method of processing media content data, the media content data comprising audio, the method including:

providing frequency domain data corresponding to said content data;

performing a log-mapping process on said frequency domain data to yield logarithmically-sampled data; and

using said logarithmically-sampled data in a process that produces an identifier associated with said media content data.

11. The method of claim 10 that further includes generating power spectrum data corresponding to said logarithmically-sampled data, and using said power spectrum data in a process that produces an identifier associated with said media content data.

12. The method of claim 10 that includes checking a database to identify content corresponding to said identifier.

13. The method of claim 10 that further comprises:

generating plural identifiers, associated with plural non-identical subsets of said media content data; and

storing said plural identifiers in a database for later matching against identifiers derived from unknown content data;

wherein the method is characterized by identifying plural non-identical subsets of data from said media content data, but storing identifiers in said database for only every Nth of said subsets.

14. The method of claim 13 that includes generating identifiers only for every Nth of said subsets.

15. In a method of processing digital audio data to yield an identifier relating to same, a subcombination of acts comprising:

providing frequency domain data corresponding to said digital audio data;

producing power spectrum data from said frequency domain data; and

performing a log-mapping process on said power spectrum data to yield logarithmically-sampled data.

16. The method of claim 15 that includes using said logarithmically-sampled data in a process that produces an identifier corresponding to said audio content.

17. The method of claim 16 that includes further processing the logarithmically-sampled data, and then matching the further processed data with one or more pre-existing signals.

18. A method of compiling a fingerprint database for identifying media content, the media content data representing at least one of audio and video, the method including—

for a particular item of known media content;

generating plural content fingerprints for plural non-identical excerpts of known media content data; and

storing said plural content fingerprints in a database for later matching against fingerprint data derived from unknown content data;

wherein the method is characterized by identifying plural non-identical subsets of data from said known media content data, and storing fingerprint data in said database for only every Nth of said subsets.

19. The method of claim 18 that includes generating content fingerprints only for every Nth of said subsets.