Compressing and decompressing weight factors using temporal prediction for audio data

• US 7,801,735 B2
• Filed: 09/25/2007
• Issued: 09/21/2010
• Est. Priority Date: 09/04/2002
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Thumpudi, Naveen, Chen, Wei-Ge
• Primary Examiner(s): Hudspeth; David R
• Assistant Examiner(s): Rider; Justin W
• Application Number:
  US11/861,122
• Publication Number:
  US 20080021704A1
• Time in Patent Office: 1,092 Days
• Field of Search: 704503-504
• US Class Current: 704/503
• 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
• CPC Class Codes: G10L 19/032 Quantisation or dequantisation of spectral components

Abstract

An audio encoder and decoder use architectures and techniques that improve the efficiency of quantization (e.g., weighting) and inverse quantization (e.g., inverse weighting) in audio coding and decoding. The described strategies include various techniques and tools, which can be used in combination or independently. For example, an audio encoder quantizes audio data in multiple channels, applying multiple channel-specific quantizer step modifiers, which give the encoder more control over balancing reconstruction quality between channels. The encoder also applies multiple quantization matrices and varies the resolution of the quantization matrices, which allows the encoder to use more resolution if overall quality is good and use less resolution if overall quality is poor. Finally, the encoder compresses one or more quantization matrices using temporal prediction to reduce the bitrate associated with the quantization matrices. An audio decoder performs corresponding inverse processing and decoding.

Claims

1. In a computing device that implements an audio encoder, a computer-implemented method comprising:

receiving, at the computing device that implements the audio encoder, audio data;

with the computing device that implements the audio encoder, encoding the audio data to produce encoded audio, including;

selecting a quantization matrix resolution from multiple available quantization matrix resolutions;

computing plural quantization matrices;

quantizing the plural quantization matrices according to the selected quantization matrix resolution; and

compressing at least one of the plural quantization matrices using temporal prediction, including, for a current weight factor of a current matrix of the plural quantization matrices;

determining a corresponding weight factor in an anchor matrix;

determining a difference between the current weight factor and the corresponding weight factor; and

entropy coding the difference between the current weight factor and the corresponding weight factor.

2. The method of claim 1 wherein the audio data is in more than two channels.

3. The method of claim 1 further comprising:

with the computing device that implements the audio encoder, decompressing the plural quantization matrices; and

with the computing device that implements the audio encoder, quantizing the audio data, including applying the plural quantization matrices.

4. The method of claim 1 further comprising, with the computing device that implements the audio encoder, outputting information for the plural compressed quantization matrices.

5. The method of claim 1 wherein the temporal prediction is from the anchor matrix to the current matrix within a channel.

6. The method of claim 1 wherein the compressing further includes performing a resampling process on the anchor matrix for temporal prediction of the current matrix with a different size than the anchor matrix.

7. The method of claim 1 wherein the compressing includes:

computing a prediction for the current matrix relative to the anchor matrix; and

computing a residual from the current matrix and the prediction.

8. In a computing device that implements an audio decoder, a computer-implemented method comprising:

receiving, at the computing device that implements the audio decoder, encoded audio data;

with the computing device that implements the audio decoder, decoding the encoded audio data, including;

selecting a quantization matrix resolution from multiple available quantization matrix resolutions;

retrieving information for plural quantization matrices; and

decompressing at least one of the plural quantization matrices using temporal prediction, including, for a current weight factor of a current matrix of the plural quantization matrices;

determining a corresponding weight factor in an anchor matrix;

entropy decoding a difference between the current weight factor and the corresponding weight factor; and

combining the corresponding weight factor with the difference between the current weight factor and the corresponding weight factor.

9. The method of claim 8 wherein the audio data is in one or more channels.

10. The method of claim 8 further comprising, with the computing device that implements the audio decoder, inverse quantizing the audio data, including applying the plural quantization matrices, wherein the decoder performs the inverse quantizing in a combined step for quantization, and wherein for each of plural coefficients the combined step includes a single multiplication by a total quantization amount.

11. The method of claim 8 wherein the temporal prediction is from the anchor matrix to the current matrix within a channel.

12. The method of claim 11 wherein the decoder resets anchor matrices at the beginning of each frame.

13. The method of claim 8 wherein the decompressing further includes performing a resampling process on the anchor matrix for temporal prediction of the current matrix with a different size than the anchor matrix.

14. The method of claim 13 wherein the size is in terms of number of bands.

15. The method of claim 8 wherein the decompressing includes:

computing a prediction for the current matrix relative to the anchor matrix;

decoding a residual for the current matrix; and

adding the residual and the prediction for the current matrix.

16. The method of claim 8 wherein the decompressing includes:

computing a prediction for the current matrix relative to the anchor matrix;

getting a bit that indicates the presence or absence of a residual for the current matrix; and

if the residual is present for the current matrix, decoding the residual and adding the residual and the prediction for the current matrix.

17. In a computing device that implements an audio encoder, a method comprising:

receiving, at the computing device that implements the audio encoder, audio;

with the computing device that implements the audio encoder, encoding the audio to produce encoded audio information, including;

selecting a weight factor resolution from multiple available weight factor resolutions;

generating plural weight factors, wherein each of the plural weight factors indicates a weight value for one or more frequency bands for a time window of the audio;

quantizing the plural weight factors according to the selected weight factor resolution;

encoding the plural quantized weight factors, including;

determining whether or not to use temporal prediction;

if using temporal prediction, for a current weight factor of the plural weight factors, the current weight factor indicating a weight value for one or more current frequency bands for a current time window;

determining a corresponding weight factor for the one or more current frequency bands for a previous time window;

determining a difference between the current weight factor and the corresponding weight factor; and

entropy coding the difference between the current weight factor and the corresponding weight factor; and

otherwise, if not using temporal prediction, for the current weight factor;

determining a previous weight factor for the one or more other frequency bands for the current time window;

determining a difference between the current weight factor and the previous weight factor; and

entropy coding the difference between the current weight factor and the previous weight factor; and

outputting, from the computing device that implements the audio encoder, the encoded audio information in a bit stream, the encoded audio information including;

information indicating the selected weight factor resolution; and

the entropy coded differences.

18. The method of claim 17 wherein the multiple available weight factor resolutions include one or more of 1 dB, 2 dB, 3 dB and 4 dB.

19. The method of claim 17 wherein the selected weight factor resolution changes over time during the encoding of the audio.

20. The method of claim 19 wherein the selection of the weight factor resolution occurs on a frame-by-frame basis.

21. The method of claim 17 wherein the current weight factor is part of a first set of weight factors for the current time window, and wherein the corresponding weight factor is part of a second set of weight factors for the previous time window.

22. The method of claim 21 wherein the first set of weight factors and the second set of weight factors have the same number of weight factors, and wherein the determining the corresponding weight factor comprises determining which weight factor in the second set is for the one or more current frequency bands.

23. The method of claim 21 wherein the first set of weight factors and the second set of weight factors have different numbers of weight factors, and wherein the determining the corresponding weight factor comprises:

mapping the one or more current frequency bands to a corresponding frequency band for the second set; and

assigning the corresponding weight factor as the weight factor in the second set for the corresponding frequency band.

24. The method of claim 17 wherein the plural weight factors include a first set of weight factors for the previous time window and a second set of weight factors for the current time window, wherein the first set of weight factors is encoded without using temporal prediction, and wherein the second set of weight factors is encoded using temporal prediction relative to the first set of weight factors.

25. The method of claim 24 wherein the first set of weight factors is also used in temporal prediction for one or more additional sets of weight factors for later time windows after the current time window.

26. In a computing device that implements an audio decoder, a method comprising:

receiving, at the computing device that implements the audio decoder, encoded audio information in a bit stream, the encoded audio information including;

information indicating a selected weight factor resolution; and

entropy coded differences for plural weight factors, wherein each of the plural weight factors indicates a weight value for one or more frequency bands for a time window of the audio;

with the computing device that implements the audio decoder, decoding the audio using the encoded audio information, including;

based at least in part upon the information indicating the selected weight factor resolution, selecting a weight factor resolution from multiple available weight factor resolutions;

decoding the plural weight factors, including;

determining whether or not to use temporal prediction;

if using temporal prediction, for a current weight factor of the plural weight factors, the current weight factor indicating a weight value for one or more current frequency bands for a current time window;

determining a corresponding weight factor for the one or more current frequency bands for a previous time window;

entropy decoding a difference between the current weight factor and the corresponding weight factor; and

combining the corresponding weight factor with the difference between the current weight factor and the corresponding weight factor; and

otherwise, if not using temporal prediction, for the current weight factor;

determining a previous weight factor for the one or more other frequency bands for the current time window;

entropy decoding a difference between the current weight factor and the previous weight factor; and

combining the previous weight factor with the difference between the current weight factor and the previous weight factor; and

inverse quantizing the plural weight factors according to the selected weight factor resolution.

27. The method of claim 26 wherein the multiple available weight factor resolutions include one or more of 1 dB, 2 dB, 3 dB and 4 dB.

28. The method of claim 26 wherein the selected weight factor resolution changes over time during the decoding of the audio.

29. The method of claim 28 wherein the selection of the weight factor resolution occurs on a frame-by-frame basis.

30. The method of claim 26 wherein the current weight factor is part of a first set of weight factors for the current time window, and wherein the corresponding weight factor is part of a second set of weight factors for the previous time window.

31. The method of claim 30 wherein the first set of weight factors and the second set of weight factors have the same number of weight factors, and wherein the determining the corresponding weight factor comprises determining which weight factor in the second set is for the one or more current frequency bands.

32. The method of claim 30 wherein the first set of weight factors and the second set of weight factors have different numbers of weight factors, and wherein the determining the corresponding weight factor comprises:

mapping the one or more current frequency bands to a corresponding frequency band for the second set; and

assigning the corresponding weight factor as the weight factor in the second set for the corresponding frequency band.

33. The method of claim 26 wherein the plural weight factors include a first set of weight factors for the previous time window and a second set of weight factors for the current time window, wherein the first set of weight factors is decoded without using temporal prediction, and wherein the second set of weight factors is decoded using temporal prediction relative to the first set of weight factors.

34. The method of claim 33 wherein the first set of weight factors is also used in temporal prediction for one or more additional sets of weight factors for later time windows after the current time window.