Method for generating an audio signature based on time domain features

• US 7,643,994 B2
• Filed: 12/02/2005
• Issued: 01/05/2010
• Est. Priority Date: 12/06/2004
• Status: Active Grant

• Current Assignee: Sony Deutschland GmbH (Sony Group Corp.)
• Original Assignee: Sony Deutschland GmbH (Sony Group Corp.)
• Inventors: Kemp, Thomas
• Primary Examiner(s): Smits; Talivaldis Ivars
• Assistant Examiner(s): BORSETTI, GREG
• Application Number:
  US11/293,072
• Publication Number:
  US 20060120536A1
• Time in Patent Office: 1,495 Days
• Field of Search: 704/243, 704/E19.009, 713/176
• US Class Current: 704/243
• CPC Class Codes: G06F 2218/08 Feature extraction
• CPC Class Codes: G10L 25/48 specially adapted for particular use

Abstract

A method for analyzing audio data (AD) is provided wherein a signature (SAD) is generated based on a feature set (FS) comprising a plurality of time domain related features (TDFk). The signature (SAD) is formed in order to represent a plurality of temporal positions or ranges of a plurality of characteristics of time domain related features (TDFk) within said audio data (AD).

Claims

1. An apparatus implementing a method of obtaining a signature for audio data, the method comprising:

receiving at the apparatus audio data to be analyzed;

determining at the apparatus a plurality of time domain features in said audio data;

selecting at the apparatus a subset of said time domain features according to their position in the order of the magnitude or strength of the respective features; and

generating at the apparatus a signature being descriptive, representative or characteristic for said audio data,

wherein said signature is exclusively based on temporal positions of said subset of said time domain features.

2. The apparatus implementing the method according to claim 1, wherein audio data of a finite length are used, in particular with a determined beginning and a determined ending.

3. The apparatus implementing the method according to claim 2, wherein said temporal positions describe absolute temporal positions of said subset of said time domain features.

4. The apparatus implementing the method according to claim 1, wherein said temporal positions describe relative temporal shifts between two successive time domain features of said subset.

5. The apparatus implementing the method according to claim 1, wherein characteristics of the same time domain feature are involved when generating said signature for said audio data.

6. The apparatus implementing the method according to claim 1, wherein characteristics of time domain features of a same type are involved when generating said signature for said audio data.

7. The apparatus implementing the method according to claim 1, wherein characteristics of time domain features of different types are involved when generating said signature for said audio data.

8. The apparatus implementing the method according to claim 1, wherein at least one of a strength, a time course of a strength and local extremer of a strength of time domain features are used as characteristics of the time domain features.

9. The apparatus implementing the method according to claim 1, wherein time domain features of a same type are combined.

10. The apparatus implementing the method according to claim 1, wherein time domain features of different types are combined.

11. The apparatus implementing the method according to claim 1, wherein an energy contour of the audio data (AD) is used as a time domain feature.

12. The apparatus implementing the method according to claim 1, wherein a zero crossing rate of the audio data is used as a time domain feature.

13. The apparatus implementing the method according to claim 1, wherein analog-to-digital converted data are used as said audio data.

14. The apparatus implementing the method according to claim 1, wherein down-sampled data are used as said audio data.

15. The apparatus implementing the method according to claim 1, wherein stereo-to-mono converted data are used as said audio data.

16. The apparatus implementing the method according to claim 1, wherein data having a frame structure are used as said audio data.

17. The apparatus implementing the method according to claim 16, wherein said frame structure is formed by a sequence of consecutive time frames, each time frame having a predetermined frame size and each time frame being shifted with respect to a succeeding time frame by a predetermined frame shift.

18. The apparatus implementing the method according to claim 17, wherein the frame shift is chosen to be less than the frame size.

19. The apparatus implementing the method according to claim 17, wherein the frame size is substantially equal to 16 ms and the frame shift is substantially equal to 10 ms.

20. The apparatus implementing the method according to claim 17, wherein said time frames are indexed and said temporal positions are given based on said index.

21. The apparatus implementing the method according to claim 1, comprising receiving, reproducing, or generating said audio data.

22. The apparatus implementing the method according to claim 1, wherein temporal positions in said signature are given in their temporal order.

23. The apparatus implementing the method according to claim 1, wherein

the time domain features of said subset of said time domain features are sorted in the order of the magnitude or strength of the respective features;

and

said signature is based on temporal positions of said sorted time domain features of said subset.

24. An apparatus implementing a method of analyzing audio data for obtaining a comparison result for the audio data, the method comprising:

receiving at the apparatus first audio data to be analyzed;

determining at the apparatus a plurality of time domain features in said first audio data;

selecting at the apparatus a subset of said time domain features according to their position in the order of the magnitude or strength of the respective features; and

generating at the apparatus a signature being descriptive, representative or characteristic for said first audio data, wherein said signature is exclusively based on temporal positions or temporal ranges of said subset of said time domain features;

providing at the apparatus a comparison signature, said comparison signature being descriptive, representative or characteristic for second audio data; and

determining at the apparatus a comparison result based on comparing said signature with said comparison signature.

25. The apparatus implementing the method according to claim 24, wherein at least two samples of audio data are compared to each other, one of said two samples being assigned to said signature and the other one of said samples being assigned to said comparison signature, wherein said comparing is based on determining coinciding features and/or differing features.

26. A computer readable storage medium including computer program instructions that cause a computer to execute a method for analyzing audio data, the method comprising:

receiving audio data to be analyzed;

determining a plurality of time domain features in said audio data;

selecting a subset of said time domain features according to their position in the order of the magnitude or strength of the respective features; and

generating a signature being descriptive, representative or characteristic for said audio data,

wherein said signature is exclusively based on temporal positions of said subset of said time domain features.