Low complexity decoder for complex transform coding of multi-channel sound

• US 8,046,214 B2
• Filed: 06/22/2007
• Issued: 10/25/2011
• Est. Priority Date: 06/22/2007
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Chen, Wei-Ge, Mehrotra, Sanjeev
• Primary Examiner(s): COLUCCI, MICHAEL C
• Application Number:
  US11/767,457
• Publication Number:
  US 20080319739A1
• Time in Patent Office: 1,586 Days
• Field of Search: 704/200.1, 704/500, 704/273, 704/216, 704/219, 704/229, 704/230, 704/246, 375/148, 375/141, 375/240, 375/240.12, 375/350, 381/63, 381/310, 341/155, 345/424, 455/200.1, 379/406.14
• US Class Current: 704/200.1
• CPC Class Codes: G10L 19/008 Multichannel audio signal coding or decoding using interchannel correlation to reduce redundancy, e.g. joint-stereo, intensity-coding or matrixing

Abstract

A multi-channel audio decoder provides a reduced complexity processing to reconstruct multi-channel audio from an encoded bitstream in which the multi-channel audio is represented as a coded subset of the channels along with a complex channel correlation matrix parameterization. The decoder translates the complex channel correlation matrix parameterization to a real transform that satisfies the magnitude of the complex channel correlation matrix. The multi-channel audio is derived from the coded subset of channels via channel extension processing using a real value effect signal and real number scaling.

Claims

1. A method of decoding multi-channel audio, the method comprising:

decoding a set of cross-channel correlation and channel power parameters from an encoded audio stream;

deriving a real number matrix transform from the set of cross-channel correlation and channel power parameters that satisfies a magnitude of cross-channel correlation;

reconstructing spectral coefficients of a coded subset of channels of the multi-channel audio;

with a processing unit, performing channel extension processing from the reconstructed spectral coefficients of the coded subset of channels based on the real number matrix transform to reconstruct spectral coefficients of the channels of the multi-channel audio; and

applying an inverse time-frequency transform to reconstruct the multi-channel audio.

2. The method of claim 1 wherein the channel extension processing comprises:

applying a real-value scaling to the coded subset of channels of the multi-channel audio;

producing a real-value effect signal using a reverb filter on at least a portion of the coded subset of channels of the multi-channel audio; and

combining a scaled version of the real-value effect signal and scaled coded subset of channels to reconstruct spectral coefficients of the channels of the multi-channel audio.

3. The method of claim 2 wherein the reverb filter is an IIR filter having real-value input and output.

4. The method of claim 1 wherein the inverse time-frequency transform is the modulated complex lapped transform.

5. The method of claim 1 wherein said reconstructing spectral coefficients of a coded subset of channels of the multi-channel audio comprises:

decoding base spectral coefficients from an encoded bitstream;

applying an inverse time-frequency transform;

applying a forward time-frequency transform;

decoding vector quantization parameters from the encoded bitstream; and

performing frequency extension processing to reconstruct the spectral coefficients of the coded subset of channels of the multi-channel audio.

6. The method of claim 1 wherein the set of cross-channel correlation and channel power parameters characterize a complex channel correlation matrix.

7. The method of claim 6 wherein the set of cross-channel correlation and channel power parameters comprise a normalized correlation matrix parameterization of the complex channel correlation matrix.

8. The method of claim 7 wherein the normalized correlation matrix parameterization comprise the parameters:

9. The method of claim 8 wherein the real number matrix is derived from the normalized correlation matrix parameterization according to the formula:

10. The method of claim 9 wherein the multi-channel audio represented in the encoded audio stream is scaled by a power-preserving scale factor by the encoder, and the method further comprises:

scaling by an inverse of the power-preserving scale factor.

11. The method of claim 10 wherein the real number matrix with said scaling by the inverse of the power-preserving scale factor is derived from the normalized correlation matrix parameterization according to the formula:

12. A method of decoding multi-channel audio, the method comprising:

decoding a set of cross-channel correlation and channel power parameters from an encoded audio stream;

deriving a real number matrix transform from the set of cross-channel correlation and channel power parameters that satisfies a magnitude of cross-channel correlation;

reconstructing spectral coefficients of a coded subset of channels of the multi-channel audio;

with a processing unit, performing channel extension processing from the reconstructed spectral coefficients of the coded subset of channels based on the real number matrix transform to reconstruct spectral coefficients of the channels of the multi-channel audio; and

applying an inverse time-frequency transform to reconstruct the multi-channel audio, wherein;

the set of cross-channel correlation and channel power parameters characterize a complex channel correlation matrix, and

the set of cross-channel correlation and channel power parameters comprise an LMRM parameterization of the complex channel correlation matrix.

13. The method of claim 12 wherein the channel extension processing comprises:

applying a real-value scaling to the coded subset of channels of the multi-channel audio;

producing a real-value effect signal using a reverb filter on at least a portion of the coded subset of channels of the multi-channel audio; and

combining a scaled version of the real-value effect signal and scaled coded subset of channels to reconstruct spectral coefficients of the channels of the multi-channel audio.

14. The method of claim 13 wherein the reverb filter is an IIR filter having real-value input and output.

15. The method of claim 12 wherein the inverse time-frequency transform is the modulated complex lapped transform.

16. The method of claim 12 wherein said reconstructing spectral coefficients of a coded subset of channels of the multi-channel audio comprises:

decoding base spectral coefficients from an encoded bitstream;

applying an inverse time-frequency transform;

applying a forward time-frequency transform;

decoding vector quantization parameters from the encoded bitstream; and

performing frequency extension processing to reconstruct the spectral coefficients of the coded subset of channels of the multi-channel audio.

17. A multi-channel audio decoder, comprising:

an input for receiving an encoded audio stream;

a processing unit configured to reconstruct multi-channel audio from the encoded audio stream via;

decoding a set of cross-channel correlation and channel power parameters from the encoded audio stream;

deriving a real number matrix transform from the set of cross-channel correlation parameters that satisfies a magnitude of cross-channel correlation;

reconstructing spectral coefficients of a coded subset of channels of the multi-channel audio;

performing channel extension processing from the reconstructed spectral coefficients of the coded subset of channels based on the real number matrix transform to reconstruct spectral coefficients of the channels of the multi-channel audio; and

applying an inverse time-frequency transform to reconstruct the multi-channel audio.

18. The multi-channel audio decoder of claim 17 wherein the set of cross-channel correlation and channel power parameters comprise a normalized correlation matrix parameterization of a complex channel correlation matrix.

19. The multi-channel audio decoder of claim 18 wherein the normalized correlation matrix parameterization comprise the parameters:

20. The multi-channel audio decoder of claim 19 wherein the real number matrix is derived from the normalized correlation matrix parameterization according to the formula:

21. The multi-channel audio decoder of claim 20 wherein the multi-channel audio represented in the encoded audio stream is scaled by a power-preserving scale factor by the encoder, and the method further comprises:

scaling by an inverse of the power-preserving scale factor.

22. The multi-channel audio decoder of claim 21 wherein the real number matrix with said scaling by the inverse of the power-preserving scale factor is derived from the normalized correlation matrix parameterization according to the formula:

23. The multi-channel audio decoder of claim 17 wherein the set of cross-channel correlation and channel power parameters characterize a complex channel correlation matrix.

24. The multi-channel audio decoder of claim 23 wherein the set of cross-channel correlation and channel power parameters comprise an LMRM parameterization of the complex channel correlation matrix.

25. The multi-channel audio decoder of claim 17, further comprising computer-readable media for providing computer-readable instructions that when executed by the processing unit, cause the processing unit to perform the acts of decoding, deriving, reconstructing, performing channel extension processing, and applying an inverse frequency transform.

26. A method of encoding multi-channel audio, the method comprising:

encoding a subset of channels of the multi-channel audio in an encoded bitstream;

with a processing unit, encoding parameters characterizing a complex channel correlation matrix in the encoded bitstream;

encoding a plurality of syntax elements for channel extension processing at decoding into the encoded bitstream, the syntax elements comprising at least the following;

a first syntax element representing a value at which to cap an effect signal for channel extension processing;

a second syntax element indicative of whether power adjustment scaling is applied;

a third syntax element representing a value at which a scale factor for channel extension processing is capped; and

a fourth syntax element indicative of which filter tap of a reverb filter generates an effect signal for channel extension processing.

27. The method of claim 26 wherein the syntax elements further comprise a fifth syntax element indicative of whether the parameters are an LMRM parameterization or a normalized power correlation matrix parameterization of the complex channel correlation matrix.

28. Computer-readable memory or storage storing computer-readable instructions that when executed by a computer cause the computer to perform a method of decoding multi-channel audio, the method comprising:

decoding a set of cross-channel correlation and channel power parameters from an encoded audio stream;

deriving a real number matrix transform from the set of cross-channel correlation and channel power parameters that satisfies a magnitude of cross-channel correlation;

reconstructing spectral coefficients of a coded subset of channels of the multi-channel audio;

performing channel extension processing from the reconstructed spectral coefficients of the coded subset of channels based on the real number matrix transform to reconstruct spectral coefficients of the channels of the multi-channel audio; and

applying an inverse time-frequency transform to reconstruct the multi-channel audio.

29. The computer-readable memory or storage of claim 28 wherein the channel extension processing comprises:

applying a real-value scaling to the coded subset of channels of the multi-channel audio;

producing a real-value effect signal using a reverb filter on at least a portion of the coded subset of channels of the multi-channel audio; and

combining a scaled version of the real-value effect signal and scaled coded subset of channels to reconstruct spectral coefficients of the channels of the multi-channel audio.

30. The computer-readable memory or storage of claim 29 wherein the reverb filter is an IIR filter having real-value input and output.

31. The computer-readable memory or storage of claim 28 wherein the inverse time-frequency transform is the modulated complex lapped transform.

32. The computer-readable memory or storage of claim 28 wherein said reconstructing spectral coefficients of a coded subset of channels of the multi-channel audio comprises:

decoding base spectral coefficients from an encoded bitstream;

applying an inverse time-frequency transform;

applying a forward time-frequency transform;

decoding vector quantization parameters from the encoded bitstream; and

performing frequency extension processing to reconstruct the spectral coefficients of the coded subset of channels of the multi-channel audio.

33. The computer-readable memory or storage of claim 28 wherein the set of cross-channel correlation and channel power parameters characterize a complex channel correlation matrix.

34. The computer-readable memory or storage of claim 33 wherein the set of cross-channel correlation and channel power parameters comprise a normalized correlation matrix parameterization of the complex channel correlation matrix.

35. The computer-readable memory or storage of claim 34 wherein the normalized correlation matrix parameterization comprise the parameters:

36. The computer-readable memory or storage of claim 35 wherein the real number matrix is derived from the normalized correlation matrix parameterization according to the formula:

37. The computer-readable memory or storage of claim 36 wherein the multi-channel audio represented in the encoded audio stream is scaled by a power-preserving scale factor by the encoder, and the method further comprises:

scaling by an inverse of the power-preserving scale factor.

38. The computer-readable memory or storage of claim 37 wherein the real number matrix with said scaling by the inverse of the power-preserving scale factor is derived from the normalized correlation matrix parameterization according to the formula:

39. Computer-readable memory or storage storing computer-readable instructions that when executed by a computer cause the computer to perform a method of decoding multi-channel audio, the method comprising:

decoding a set of cross-channel correlation and channel power parameters from an encoded audio stream;

deriving a real number matrix transform from the set of cross-channel correlation and channel power parameters that satisfies a magnitude of cross-channel correlation;

reconstructing spectral coefficients of a coded subset of channels of the multi-channel audio;

performing channel extension processing from the reconstructed spectral coefficients of the coded subset of channels based on the real number matrix transform to reconstruct spectral coefficients of the channels of the multi-channel audio; and

applying an inverse time-frequency transform to reconstruct the multi-channel audio, wherein;

the set of cross-channel correlation and channel power parameters characterize a complex channel correlation matrix, and

the set of cross-channel correlation and channel power parameters comprise an LMRM parameterization of the complex channel correlation matrix.

40. The computer-readable memory or storage of claim 39 wherein the channel extension processing comprises:

applying a real-value scaling to the coded subset of channels of the multi-channel audio;

producing a real-value effect signal using a reverb filter on at least a portion of the coded subset of channels of the multi-channel audio; and

combining a scaled version of the real-value effect signal and scaled coded subset of channels to reconstruct spectral coefficients of the channels of the multi-channel audio.

41. The computer-readable memory or storage of claim 40 wherein the reverb filter is an IIR filter having real-value input and output.

42. The computer-readable memory or storage of claim 39 wherein the inverse time-frequency transform is the modulated complex lapped transform.

43. The computer-readable memory or storage of claim 39 wherein said reconstructing spectral coefficients of a coded subset of channels of the multi-channel audio comprises:

decoding base spectral coefficients from an encoded bitstream;

applying an inverse time-frequency transform;

applying a forward time-frequency transform;

decoding vector quantization parameters from the encoded bitstream; and

performing frequency extension processing to reconstruct the spectral coefficients of the coded subset of channels of the multi-channel audio.

44. Computer-readable memory or storage storing computer-readable instructions that when executed by a computer cause the computer to perform a method of encoding multi-channel audio, the method comprising:

encoding a subset of channels of the multi-channel audio in an encoded bitstream;

encoding parameters characterizing a complex channel correlation matrix in the encoded bitstream;

encoding a plurality of syntax elements for channel extension processing at decoding into the encoded bitstream, the syntax elements comprising at least the following;

a first syntax element representing a value at which to cap an effect signal for channel extension processing;

a second syntax element indicative of whether power adjustment scaling is applied;

a third syntax element representing a value at which a scale factor for channel extension processing is capped; and

a fourth syntax element indicative of which filter tap of a reverb filter generates an effect signal for channel extension processing.

45. The computer-readable memory or storage of claim 44 wherein the syntax elements further comprise a fifth syntax element indicative of whether the parameters are an LMRM parameterization or a normalized power correlation matrix parameterization of the complex channel correlation matrix.

46. A multi-channel audio encoder, comprising:

an output for transmitting the encoded bitstream;

a processing unit; and

the computer-readable memory or storage of claim 44, wherein the processing unit is operable to execute the computer-readable instructions to encode the multi-channel audio as the encoded bitstream.