Device and method for analyzing an information signal

• US 20090265024A1
• Filed: 06/30/2009
• Published: 10/22/2009
• Est. Priority Date: 05/07/2004
• Status: Active Grant

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Gracenote, Inc. (RR Donnelley & Sons Company)
• Inventors: Dittmar, Christian, Herre, Jurgen, Uhle, Christean
• Granted Patent: US 8,175,730 B2
• Time in Patent Office: Days
• US Class Current: 700/94
• CPC Class Codes: G10L 25/48 specially adapted for particular use

Abstract

In order to analyze an information signal, a significant short-time spectrum is extracted from the information signal, the means for extracting being configured to extract such short-time spectra which come closer to a specific characteristic than other short-time spectra of the information signal. The short-time spectra extracted are then decomposed into component signals using ICA analysis, a component signal spectrum representing a profile spectrum of a tone source which generates a tone corresponding to the characteristic sought for. From a sequence of short-time spectra of the information signal and from the profile spectra determined, an amplitude envelope is eventually calculated for each profile spectrum, the amplitude envelope indicating how a profile spectrum of a tone source all in all changes over time. The profile spectra and all the amplitude envelopes associated therewith provide a description of the information signal which may be evaluated further, for example for transcription purposes in the case of a music signal.

Claims

1. A device, comprising:

an extractor to provide extracted short-time spectra by extracting short-time spectra or derived short-time spectra having at least one of harmonic or percussive portions from an information signal;

a decomposer to decompose the extracted short-time spectra into component signal spectra representing profile spectra for a plurality of tone sources, the profile spectra determined in part by a reduced number of the extracted short-time spectra resulting from a weighted linear combination of the extracted short-time spectra; and

a calculator to calculate a plurality of amplitude envelopes over time on the basis of the profile spectra and the extracted short-time spectra, the plurality of amplitude envelopes corresponding to the plurality of tone sources.

2. The device of claim 1, wherein the extractor further comprises:

at least one high-pass filter.

3. The device of claim 1, wherein the extractor further comprises:

a differentiator.

4. The device of claim 1, wherein the extractor further comprises:

a maximum searcher.

5. The device of claim 4, wherein the maximum searcher is to receive input comprising phase information derived from the information signal.

6. The device of claim 1, wherein the extractor is to implement a smoothed summation of the extracted short-time spectra to provide a detection function over time.

7. The device of claim 1, wherein the decomposer is to perform a principal component analysis.

8. The device of claim 1, wherein the decomposer is to perform an independent component analysis.

9. The device of claim 1, further comprising:

a classifier to classify the component signal spectra into percussive component signals and non-percussive component signals based on at least one of the amplitude envelopes or the profile spectra.

10. A method, comprising:

extracting short-time spectra or derived short-time spectra having at least one of harmonic or percussive portions from an information signal to provide extracted short-time spectra;

decomposing the extracted short-time spectra into component signal spectra representing profile spectra for a plurality of tone sources, the profile spectra determined in part by a reduced number of the extracted short-time spectra resulting from a weighted linear combination of the extracted short-time spectra; and

calculating a plurality of amplitude envelopes over time on the basis of the profile spectra and the extracted short-time spectra, the plurality of amplitude envelopes corresponding to the plurality of tone sources.

11. The method of claim 10, comprising:

transforming the information signal into at least one of an amplitude or a phase spectrogram.

12. The method of claim 11, wherein the transforming is accomplished using a Fourier transform and a selected hopping period.

13. The method of claim 11, wherein the extracting further comprises:

differentiation along a temporal expansion of the amplitude spectrogram.

14. The method of claim 10, wherein the decomposing further comprises:

performing a principal component analysis on the extracted short-time spectra.

15. The method of claim 10, wherein the decomposing further comprises:

decorrelating the extracted short-time spectra.

16. The method of claim 10, wherein the decomposing further comprises:

normalizing the extracted short-time spectra.

17. The method of claim 10, wherein the decomposing further comprises:

performing an independent component analysis on the extracted short-time spectra.

18. The method of claim 10, comprising:

classifying the profile spectra into percussive and non-percussive subsets.

19. The method of claim 10, comprising:

comparing a feature extracted from the profile spectra or the amplitude envelopes with features of known sources stored in a database to classify at least one of the known sources

20. A tangible computer storage medium having stored thereon a computer program which, when executed by a computer, results in the computer performing a method comprising:

extracting short-time spectra or derived short-time spectra having at least one of harmonic or percussive portions from an information signal to provide extracted short-time spectra;

decomposing the extracted short-time spectra into component signal spectra representing profile spectra for a plurality of tone sources, the profile spectra determined in part by a reduced number of the extracted short-time spectra resulting from a weighted linear combination of the extracted short-time spectra; and

calculating a plurality of amplitude envelopes over time on the basis of the profile spectra and the extracted short-time spectra, the plurality of amplitude envelopes corresponding to the plurality of tone sources.