Apparatus and method for characterizing an information signal

• US 20050038635A1
• Filed: 09/23/2004
• Published: 02/17/2005
• Est. Priority Date: 07/19/2002
• Status: Active Grant

• Current Assignee: Callahan Cellular LLC (Intellectual Ventures LLC)
• Original Assignee: Fraunhofer Gesellschaft Zur Foerderung Der Angewandten Forsching E.V.
• Inventors: Klefenz, Frank, Brandenburg, Karlheinz
• Granted Patent: US 7,035,742 B2
• Time in Patent Office: Days
• US Class Current: 702/189
• CPC Class Codes: G10H 2240/135 Library retrieval index, i.e. using an indexing scheme to efficiently retrieve a music piece
• CPC Class Codes: G10H 3/125 Extracting or recognising the pitch or fundamental frequency of the picked up signal
• CPC Class Codes: G10L 15/02 Feature extraction for speech recognition; Selection of recognition unit
• CPC Class Codes: G10L 15/10 using distance or distortion measures between unknown speech and reference templates

Abstract

For characterizing an information signal having an amplitude-time waveform with local extreme values, at first the local extreme values of the information signal are determined, wherein a local extreme value is defined by a time instant and an amplitude. Furthermore, area information of valleys or mountains of the information signal in case of a one-dimensional amplitude of the information signal or volume information in case of a two-dimensional amplitude of the information signal of valleys or mountains is ascertained. A valley or mountain is defined by a temporal section of the information signal, wherein the section of the information signal extends from the time instant of a local extreme value to a temporarily adjacent value of the information signal having the same amplitude as the local extreme value. Area or volume information of several mountains or valleys is characteristic for the information signal and permits further characterization of the information signal, build-up of an information signal database, or identification of an information signal on the basis of an existing information signal database. Area or volume information is on the one hand characteristic for the information signal and on the other hand, due to its integral nature, robust against information signal changes in form of overlays or distortions.

Claims

1. An apparatus for characterizing an information signal having an amplitude-time waveform with local extreme values, comprising:

a processor for determining the local extreme values of the information signal, wherein a local extreme value is defined by a time instant and an amplitude; and

a calculator for ascertaining area information of valleys or mountains of the information signal in case of a one-dimensional amplitude of the information signal or of volume information of valleys or mountains of the information signal in case of a two-dimensional amplitude of the information signal, wherein a valley or mountain is defined by a temporal section of the information signal, wherein the section of the information signal extends from the time instant of a local extreme value to a temporarily adjacent value of the information signal having the same amplitude as the local extreme value, wherein the area information or volume information of several mountains or valleys is characteristic for the information signal.

2. The apparatus of claim 1,

wherein the processor is formed to determine local maximums of the information signal, and

wherein the calculator is formed to ascertain area information of valleys of the information signal, wherein a valley is defined by a temporal section of the information signal extending from the time instant of the local maximum to a temporarily adjacent value of the information signal having the same amplitude as the local maximum.

3. The apparatus of claim 2,

wherein the calculator is formed to determine area information about a valley of the information signal associated with a local maximum, which is present above a valley of the information signal with which a further local maximum with a smaller amplitude than the one local maximum associated, so that the area information of the one valley together with the area information of the other valley points to a geometrical area between the information signal and an upper boundary line of the one valley.

4. The apparatus of claim 1,

wherein the processor is formed to determine local minima of the information signal, and

wherein the calculator is formed to ascertain area information from mountains of the information signal, wherein a mountain is defined by a temporal section of the information signal extending from the time instant of the local minimum to a temporarily adjacent value of the information signal having the same amplitude as the local minimum.

5. The apparatus of claim 4,

wherein the calculator is formed to determine area information about a mountain of the information signal associated with a first minimum, which is present above a mountain of the information signal with which a second minimum with a smaller amplitude than the first minimum is associated, so that the area information of the first mountain together with the area information of the second mountain points to a geometric area between the information signal and a lower boundary line of the first mountain.

6. The apparatus of claim 1,

wherein the calculator is formed to calculate as area information the geometric area of a valley or mountain associated with a local extreme value.

7. The apparatus of claim 1,

wherein the calculator is formed to assign a temporal value taken from the temporal section by which the valley or the mountain is defined to the area information for a valley or a mountain, whereby a series of area information time value tupels arises.

8. The apparatus of claim 7,

wherein the time value is the time instant of the local extreme value associated with the valley or the mountain.

9. The apparatus of claim 1,

wherein the calculator is formed to ascertain quantitative values as area information.

10. The apparatus of claim 9, further comprising:

a post-processor for post-processing the area information of several mountains or valleys.

11. The apparatus of claim 10,

wherein the post-processor is formed to suppress mountains or valleys whose area information is smaller than a predetermined threshold value in a further processing.

12. The apparatus of claim 10,

wherein the post-processor is formed to ascertain a series of area information changes by determining a ratio of temporarily successive area information each or by forming a difference between two successive pieces of area information each.

13. The apparatus of claim 10,

wherein the post-processor is formed to examine the series of area information for repeating sequences of area information.

14. The apparatus of claim 13,

wherein the post-processor is formed to segment the information signal according to repeating sequences.

15. The apparatus of claim 10,

wherein the post-processor is formed to determine combination pairs of identical area information or of area information differing by less than a predetermined tolerance, and to calculate frequency information on the information signal on the basis of a time difference between time instants of area information of a combination pair.

16. The apparatus of claim 10,

wherein the post-processor is formed to determine abrupt area information changes in the series of area information, and to segment the information signal at abrupt area information changes.

17. The apparatus of claim 10,

wherein the post-processor is formed to normalize area information to obtain a series of normalized area information.

18. The apparatus of claim 10,

wherein the post-processor is formed to obtain a longitudinal extension of mountains or valleys using the area information and associated time instants, to be able to determine whether the information signal is an audio signal with voice or instrumental music.

19. The apparatus of claim 1,

wherein the calculator is formed to ascertain a series of area information or volume information,

wherein the apparatus further comprises a normalizer formed to calculate an overall area or an overall volume by addition of at least two elements of the series of area information or volume information, and further formed to divide the elements of the series by the overall area or the overall volume to obtain a normalized series.

20. The apparatus of claim 19,

wherein the normalizer is formed to totalize all elements of a segment of the series, wherein a segment represents an original pattern repeating in an unsegmented series.

21. A method for characterizing an information signal having an amplitude-time waveform with local extreme values, comprising:

determining the local extreme values of the information signal, wherein a local extreme value is defined by a time instant and an amplitude; and

ascertaining area information of valleys or mountains of the information signal in case of a one-dimensional amplitude of the information signal or volume information of valleys or mountains of the information signal in case of a two-dimensional amplitude of the information signal, wherein a valley or mountain is defined by a temporal section of the information signal, wherein the section of the information signal extends from the time instant of a local extreme value to a temporarily adjacent value of the information signal having the same amplitude as the local extreme value, wherein the area information or volume information of several mountains or valleys are characteristic for the information signal.

22. An apparatus for building up an information signal database, comprising:

a characterizer for characterizing an information signal having an amplitude-time waveform with local extreme values to obtain area information for an information signal, comprising;

a processor for determining the local extreme values of the information signal, wherein a local extreme value is defined by a time instant and an amplitude; and

a calculator for ascertaining area information of valleys or mountains of the information signal in case of a one-dimensional amplitude of the information signal or of volume information of valleys or mountains of the information signal in case of a two-dimensional amplitude of the information signal, wherein a valley or mountain is defined by a temporal section of the information signal, wherein the section of the information signal extends from the time instant of a local extreme value to a temporarily adjacent value of the information signal having the same amplitude as the local extreme value, wherein the area information or volume information of several mountains or valleys is characteristic for the information signal;

a feeder for feeding several information signals to the characterizer, to obtain, for each information signal, a feature vector derived from the area information for the information signal; and

a storage for storing feature vectors for the information signals, wherein an identifier of the information signal from which the respective feature vector has been ascertained, is associated with each feature vector.

23. A method for building up an information signal database, comprising:

characterizing an information signal having an amplitude-time waveform with local extreme values to obtain area information for an information signal, comprising;

determining the local extreme values of the information signal, wherein a local extreme value is defined by a time instant and an amplitude; and

ascertaining area information of valleys or mountains of the information signal in case of a one-dimensional amplitude of the information signal or volume information of valleys or mountains of the information signal in case of a two-dimensional amplitude of the information signal, wherein a valley or mountain is defined by a temporal section of the information signal, wherein the section of the information signal extends from the time instant of a local extreme value to a temporarily adjacent value of the information signal having the same amplitude as the local extreme value, wherein the area information or volume information of several mountains or valleys are characteristic for the information signal;

performing the step of characterizing with several information signals to obtain a feature vector derived from the area information for the information signal for each information signal; and

storing feature vectors for the information signal, wherein an identifier of the information signal from which the respective feature vector has been ascertained is associated with each feature vector.

24. An apparatus for identifying an information signal using an information signal database storing the feature vectors in association with respective information signal identifiers, comprising:

a characterizer for characterizing an information signal having an amplitude-time waveform with local extreme values, to obtain area information of several mountains or valleys, which are characteristic for the information signal, comprising;

a processor for determining the local extreme values of the information signal, wherein a local extreme value is defined by a time instant and an amplitude; and

a calculator for ascertaining area information of valleys or mountains of the information signal in case of a one-dimensional amplitude of the information signal or of volume information of valleys or mountains of the information signal in case of a two-dimensional amplitude of the information signal, wherein a valley or mountain is defined by a temporal section of the information signal, wherein the section of the information signal extends from the time instant of a local extreme value to a temporarily adjacent value of the information signal having the same amplitude as the local extreme value, wherein the area information or volume information of several mountains or valleys is characteristic for the information signal; and

a comparator for comparing a feature vector derived from the area information or volume information characteristic for a test information signal with feature vectors stored in the information signal database, to make a statement on the information signal.

25. A method for identifying an information signal using an information signal database storing the feature vectors in association with respective information signal identifiers, comprising:

characterizing an information signal having an amplitude-time waveform with local extreme values, to obtain area information of several mountains or valleys, which are characteristic for the information signal, comprising;

determining the local extreme values of the information signal, wherein a local extreme value is defined by a time instant and an amplitude; and

ascertaining area information of valleys or mountains of the information signal in case of a one-dimensional amplitude of the information signal or volume information of valleys or mountains of the information signal in case of a two-dimensional amplitude of the information signal, wherein a valley or mountain is defined by a temporal section of the information signal, wherein the section of the information signal extends from the time instant of a local extreme value to a temporarily adjacent value of the information signal having the same amplitude as the local extreme value, wherein the area information or volume information of several mountains or valleys are characteristic for the information signal; and

comparing a feature vector derived from the area information or volume information characteristic for a test information signal with feature vectors stored in the information signal database, to make a statement on the information signal.

26. A computer program with a program code for performing, when the program runs on a computer, a method for characterizing an information signal having an amplitude-time waveform with local extreme values, comprising:

determining the local extreme values of the information signal, wherein a local extreme value is defined by a time instant and an amplitude; and

ascertaining area information of valleys or mountains of the information signal in case of a one-dimensional amplitude of the information signal or volume information of valleys or mountains of the information signal in case of a two-dimensional amplitude of the information signal, wherein a valley or mountain is defined by a temporal section of the information signal, wherein the section of the information signal extends from the time instant of a local extreme value to a temporarily adjacent value of the information signal having the same amplitude as the local extreme value, wherein the area information or volume information of several mountains or valleys are characteristic for the information signal.

27. A computer program with a program code for performing, when the program runs on a computer, a method for building up an information signal database, comprising:

characterizing an information signal having an amplitude-time waveform with local extreme values to obtain area information for an information signal, comprising;

determining the local extreme values of the information signal, wherein a local extreme value is defined by a time instant and an amplitude; and

ascertaining area information of valleys or mountains of the information signal in case of a one-dimensional amplitude of the information signal or volume information of valleys or mountains of the information signal in case of a two-dimensional amplitude of the information signal, wherein a valley or mountain is defined by a temporal section of the information signal, wherein the section of the information signal extends from the time instant of a local extreme value to a temporarily adjacent value of the information signal having the same amplitude as the local extreme value, wherein the area information or volume information of several mountains or valleys are characteristic for the information signal;

performing the step of characterizing with several information signals to obtain a feature vector derived from the area information for the information signal for each information signal; and

storing feature vectors for the information signal, wherein an identifier of the information signal from which the respective feature vector has been ascertained is associated with each feature vector.

28. A computer program with a program code for performing, when the program runs on a computer, a method for identifying an information signal using an information signal database storing the feature vectors in association with respective information signal identifiers, comprising:

characterizing an information signal having an amplitude-time waveform with local extreme values, to obtain area information of several mountains or valleys, which are characteristic for the information signal, comprising;

determining the local extreme values of the information signal, wherein a local extreme value is defined by a time instant and an amplitude; and

ascertaining area information of valleys or mountains of the information signal in case of a one-dimensional amplitude of the information signal or volume information of valleys or mountains of the information signal in case of a two-dimensional amplitude of the information signal, wherein a valley or mountain is defined by a temporal section of the information signal, wherein the section of the information signal extends from the time instant of a local extreme value to a temporarily adjacent value of the information signal having the same amplitude as the local extreme value, wherein the area information or volume information of several mountains or valleys are characteristic for the information signal; and

comparing a feature vector derived from the area information or volume information characteristic for a test information signal with feature vectors stored in the information signal database, to make a statement on the information signal.