Comparison of data signals using characteristic electronic thumbprints

• US 7,421,376 B1
• Filed: 04/24/2002
• Issued: 09/02/2008
• Est. Priority Date: 04/24/2001
• Status: Active Grant

• Current Assignee: R2 Solutions LLC (Acacia Research Corporation)
• Original Assignee: Auditude Incorporated (Adobe Inc.)
• Inventors: Caruso, Jeffrey L., Yeager, William S., Seet, Nicholas
• Primary Examiner(s): Barlow; John E.
• Assistant Examiner(s): Kundu; Sujoy K
• Application Number:
  US10/132,091
• Time in Patent Office: 2,323 Days
• Field of Search: 702189-190, 341/155, 704/205, 704/222, 704/229
• US Class Current: 702/190
• CPC Class Codes: G06F 16/634 Query by example, e.g. query by humming
• CPC Class Codes: G06F 16/683 using metadata automatically derived from the content
• CPC Class Codes: G10H 2210/031 Musical analysis, i.e. isolation, extraction or identification of musical elements or musical parameters from a raw acoustic signal or from an encoded audio signal neural networks for electrophonic musical instruments or musical processing G10H2250/311
• CPC Class Codes: G10H 2240/141 Library retrieval matching, i.e. any of the steps of matching an inputted segment or phrase with musical database contents, e.g. query by humming, singing or playing; the steps may include, e.g. musical analysis of the input, musical feature extraction, query formulation, or details of the retrieval process

Abstract

A characteristic thumbprint is extracted from a data signal, the thumbprint based on statistics relating to the data signal. The data signal can be compared indirectly by matching this thumbprint against one or more reference thumbprints. The data signal may be any type of signal, including streaming digitized audio or obtained from static files. A database for containing a number of such thumbprints and methods for searching the database are also described.

Claims

1. A method for extracting a digital thumbprint based on at least a portion of a data signal, the method comprising:

filtering the data signal into a plurality of parallel filtered signals, at least two of the filtered signals having a different frequency range;

computing a set of power values for segments of each of the filtered signals, where a power value represents a measure of power contained within the segment of the filtered signal; and

creating a digital thumbprint stored on a computer readable medium, the thumbprint comprising statistical data for the power values of the filtered signals.

2. The method of claim 1, further comprising:

formatting the data signal before filtering.

3. The method of claim 2, wherein formatting includes one or more of:

decompressing the data signal;

normalizing the amplitude of the data signal; and

upsampling or downsampling the data signal.

4. The method of claim 1, wherein the plurality of filtered signals have non-overlapping frequency ranges that are chosen in a logarithmically even fashion.

5. The method of claim 1, wherein the power values comprise the root-mean-square amplitudes for each of a series of segments, a segment being a series of samples of a particular filtered signal.

6. The method of claim 1, wherein the statistical data corresponding to each filtered signal include a value related to the ratio of the standard deviation to the mean of the power values computed for the filtered signal.

7. The method of claim 1, wherein the thumbprint comprises a vector having a dimension for each filtered signal, the vector having a coordinate in each dimension that is based on the ratio of the standard deviation to the mean of the power values computed for the corresponding filtered signal.

8. The method of claim 1, wherein each power value is based on the root-mean-square amplitude of a segment of the corresponding filtered signal.

9. The method of claim 1, wherein creating the thumbprint includes normalizing the statistical data.

10. The method of claim 1, wherein the data signal is a derived from an audio file.

11. The method of claim 1, wherein the data signal is a streaming media broadcast.

12. The method of claim 1, wherein the data signal is a digitized media broadcast.

13. A method for extracting a characteristic thumbprint for a data signal, the method comprising:

receiving a series of samples representing the data signal in digital form;

filtering the data signal into a plurality of parallel filtered signals, at least some of the filtered signals having different frequency bands;

measuring the power contained within at least a portion of the filtered signals without using a Fast Fourier Transform;

computing statistical information relating to the measured power of the filtered signals; and

formatting the statistical information to create a thumbprint that is characteristic of the data signal, the thumbprint stored on a computer readable medium.

14. The method of claim 13, wherein the thumbprint comprises a vector having at least one coordinate for each filtered signal, the coordinates based on the statistical information for their corresponding filtered signal.

15. The method of claim 13, wherein the data signal represents and audio signal, the method further comprising:

before filtering, formatting the audio signal to have a predetermined amplitude scale and sampling rate.

16. The method of claim 13, wherein the audio signal is derived from a compressed audio file, the method further comprising:

decompressing the audio file.

17. The method of claim 13, wherein measuring the power of each filtered signal comprises computing the root-mean-square amplitude of a plurality of segments of the filtered signal.

18. A method for creating a thumbprint characteristic of a data signal, the method comprising:

filtering the data signal into a plurality of parallel filtered signals, at least some of the filtered signals having different frequency ranges;

a step for computing statistical information relating to a measured power of the filtered signals; and

formatting the statistical information to create a thumbprint that is characteristic of the data signal, the thumbprint stored on a computer readable medium.

19. A system for extracting a characteristic thumbprint from a data signal, the system comprising:

a plurality of filters for filtering the data signal into a plurality of parallel filtered signals, at least some of the filtered signals having different frequency bands;

a power measurement module coupled to the filters, the power measurement module adapted to determine the power contained within at least a portion of each filtered signal;

a computation module coupled to the power measurement module, the computation module adapted to compute statistical information based on the measured power of the filtered signals; and

a thumbprint generator adapted to create a thumbprint using the statistical information from the computation module.

20. The system of claim 19, wherein:

the power measurement module computes the root-mean-square amplitude of a plurality of segments for each filtered signal; and

the statistical information for each filtered signal is, at least in part, based on the ratio of the standard deviation to the mean of the root-mean-square amplitudes for the corresponding filtered signal.

21. The system of claim 19, wherein the system does not perform an FFT transformation on the data signal.

22. A system for extracting a characteristic thumbprint from a data signal, the system comprising:

means for filtering the data signal into a plurality of parallel filtered signals, at least some of the filtered signals having different frequency ranges;

means for computing statistical information relating to a measured power contained within at least a portion of the filtered signals; and

means for formatting the statistical information to create a thumbprint that is characteristic of the data signal.

23. A digital thumbprint that is characteristic of a data signal and stored on a computer readable medium, the thumbprint comprising a plurality of coordinates, wherein at least some of the coordinates are obtained by filtering that data signal into a plurality parallel filtered signals, at least two of the filtered signals having a different frequency range;

computing a set of power values for segments of each of the filtered signals, where a power value represents a measure of power contained within the segment of the filtered signal; and

computing the coordinates based on statistical data for power values of the filtered signals.

24. The thumbprint of claim 23, wherein the statistical information corresponding to a filtered signal is based on the ratio of the standard deviation to the mean of the power of the corresponding filtered signal.

25. A database of thumbprints stored on a computer-readable medium, each thumbprint as claimed in claim 23, wherein the thumbprints are organized in a tree structure indexed by a coordinate.