Quality improvement techniques in an audio encoder

• US 8,554,569 B2
• Filed: 08/27/2009
• Issued: 10/08/2013
• Est. Priority Date: 12/14/2001
• Status: Expired due to Term

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Chen, Wei-Ge, Thumpudi, Naveen, Lee, Ming-Chieh
• Primary Examiner(s): Desir, Pierre-Louis
• Assistant Examiner(s): Serrou, Abdelali
• Application Number:
  US12/549,210
• Publication Number:
  US 20090326962A1
• Time in Patent Office: 1,503 Days
• Field of Search: 704500-504, 704/223, 704/200.1, 704/258, 704/E17.001, 704/201, 704/229, 704/230, 704/E21.001
• US Class Current: 704/500
• CPC Class Codes: G10L 19/002 Dynamic bit allocation for perceptual audio coders G10L19/032
• 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 19/02 using spectral analysis, e.g. transform vocoders or subband vocoders

Abstract

An audio encoder implements multi-channel coding decision, band truncation, multi-channel rematrixing, and header reduction techniques to improve quality and coding efficiency. In the multi-channel coding decision technique, the audio encoder dynamically selects between joint and independent coding of a multi-channel audio signal via an open-loop decision based upon (a) energy separation between the coding channels, and (b) the disparity between excitation patterns of the separate input channels. In the band truncation technique, the audio encoder performs open-loop band truncation at a cut-off frequency based on a target perceptual quality measure. In multi-channel rematrixing technique, the audio encoder suppresses certain coefficients of a difference channel by scaling according to a scale factor, which is based on current average levels of perceptual quality, current rate control buffer fullness, coding mode, and the amount of channel separation in the source. In the header reduction technique, the audio encoder selectively modifies the quantization step size of zeroed quantization bands so as to encode in fewer frame header bits.

Claims

1. In an audio encoder, a method comprising:

receiving, at the audio encoder, audio in multiple channels;

with the audio encoder, encoding the audio to produce encoded audio information, including;

encoding the audio in a first set of one or more spectral bands as quantized spectral information, including;

truncating the audio in a second set of one or more spectral bands higher in frequency than the first set of one or more spectral bands, leaving the audio in the first set of one or more spectral bands;

selectively performing a multi-channel transform between the multiple channels for the audio in the first set of one or more spectral bands;

performing perceptual weighting for the audio in the first set of one or more spectral bands;

performing entropy encoding for the audio in the first set of one or more spectral bands;

encoding the audio in the second set of one or more spectral bands as parameters instead of quantized spectral information, wherein the parameters at least in part indicate forms of patterns to be generated during decoding to represent the audio in the second set of one or more spectral bands, the patterns that represent the audio in the second set of one or more spectral bands to be combined with results of decoding the quantized spectral information for the audio in the first set of one or more spectral bands, and wherein the encoding the audio in the second set of one or more spectral bands comprises;

when the multiple channels are independently coded, using a different array of noise parameters for each of the multiple independently coded channels, wherein the different array of noise parameters for each of the multiple independently coded channels includes one or more noise parameters, each of the one or more noise parameters indicating a noise parameter value for a frequency band of one or more of the spectral bands in the second set over a time window of the independently coded channel; and

when the multiple channels are jointly coded, using an array of noise parameters for the joint coding channel, wherein the array of noise parameters for the joint coding channel includes one or more noise parameters, each of the one or more noise parameters indicating a noise parameter value for a frequency band of one or more of the spectral bands in the second set over a time window of the joint coding channel; and

outputting, from the audio encoder, the encoded audio information in a bit stream.

2. The method of claim 1 wherein the truncation includes dropping spectral coefficients in the second set of one or more spectral bands after a windowed overlapped frequency transform during the encoding of the audio in the first set of one or more spectral bands.

3. The method of claim 1 wherein the encoded audio information includes, for a frame of the audio in multiple channels:

information that indicates the second set of one or more spectral bands are encoded as the parameters instead of quantized spectral information.

4. The method of claim 3 wherein the parameters and the information that indicates the second set of one or more spectral bands change on a frame-by-frame basis.

5. The method of claim 1 wherein the second set of one or more spectral bands are high bands above a threshold and the first set of one or more spectral bands are low bands below the threshold.

6. The method of claim 1 wherein the perceptual weighting of the audio in the first set of one or more spectral bands accounts for the truncation of the audio in the second set of one or more spectral bands.

7. The method of claim 1 wherein the encoding the audio in the second set of one or more spectral bands further comprises:

mapping the second set of one or more spectral bands to positions of the frequency bands for the noise parameters, respectively.

8. The method of claim 1 further comprising, as part of the encoding the audio in the first set of one or more spectral bands, scalar quantizing values of the audio after the perceptual weighting and before the entropy encoding.

14. The method of claim 1 wherein the method further comprises identifying a cutoff frequency between the first set of spectral bands and the second set of spectral bands based on perceptual audio quality for the audio.

15. The method of claim 14 wherein the perceptual audio quality is measured in terms of noise to excitation ratio.

16. The method of claim 14 wherein the perceptual audio quality is measured in terms of noise to mask ratio.

17. The method of claim 1 wherein the truncating the audio comprises:

performing first band truncation on the audio at a first cut-off frequency based on a target audio quality; and

performing second band truncation on the audio at a second cut-off frequency based on achieved audio quality after encoding of the audio after the first band truncation.

9. In an audio decoder, a method comprising:

receiving, at the audio decoder, encoded audio information for audio in multiple channels;

with the audio decoder, decoding the encoded audio information to produce reconstructed audio, including;

decoding the audio in a first set of one or more spectral bands, the audio in the first set of one or more spectral bands having been encoded as quantized spectral information, wherein the decoding the audio in the first set of one or more spectral bands includes;

performing entropy decoding for the audio in the first set of one or more spectral bands;

performing inverse quantization for the audio in the first set of one or more spectral bands; and

selectively performing an inverse multi-channel transform between the multiple channels of the audio in the first set of one or more spectral bands;

decoding the audio in a second set of one or more spectral bands higher in frequency than the first set of one or more spectral bands, the audio in the second set of one or more spectral bands having been truncated during encoding to leave the audio in the first set of one or more spectral bands, wherein the decoding the audio in the second set of one or more spectral bands uses a different array of noise parameters for each independently coded channel when the multiple channels are independently coded, wherein the decoding the audio in the second set of one or more spectral bands uses an array of noise parameters for a joint coding channel when the multiple channels are jointly coded, and wherein the decoding the audio in the second set of one or more spectral bands includes;

when the multiple channels are independently coded, for each of the multiple independently coded channels, decoding one or more noise parameters of the different array of noise parameters for the independently coded channel, each of the one or more noise parameters indicating a noise parameter value for a frequency band of one or more of the spectral bands in the second set over a time window of the independently coded channel;

when the multiple channels are jointly coded, decoding one or more noise parameters of the array of noise parameters for the joint coding channel, each of the one or more noise parameters indicating a noise parameter value for a frequency band of one or more of the spectral bands in the second set over a time window of the joint coding channel; and

based at least in part upon the noise parameters, generating patterns that represent the audio in the second set of spectral bands;

combining the patterns that represent the audio in the second set of one or more spectral bands with results of decoding the quantized spectral information for the audio in the first set of one or more spectral bands; and

outputting, from the audio decoder, the reconstructed audio for playback.

10. The method of claim 9 wherein the truncation includes any suppression or elimination of the audio for the second set of one or more spectral bands during encoding of the audio in the first set of one or more spectral bands.

11. The method of claim 9 wherein the encoded audio information includes, for a frame of the audio in multiple channels:

information that indicates the second set of one or more spectral bands are encoded as parameters instead of quantized spectral information.

12. The method of claim 11 wherein the parameters and the information that indicates the second set of one or more spectral bands change on a frame-by-frame basis.

13. The method of claim 9 wherein the decoding the audio in the second set of one or more spectral bands further comprises:

mapping the second set of one or more spectral bands to positions of the frequency bands for the noise parameters, respectively.

18. One or more computer-readable media storing instructions for causing a processing unit programmed thereby to perform a method of audio decoding, the one or more computer-readable media being selected from a group consisting of volatile memory, non-volatile memory, magnetic storage media and optical storage media, the method comprising:

receiving encoded audio information for audio in multiple channels;

decoding the encoded audio information to produce reconstructed audio, including;

decoding the audio in a first set of one or more spectral bands, the audio in the first set of one or more spectral bands having been encoded as quantized spectral information, wherein the decoding the audio in the first set of one or more spectral bands includes;

performing entropy decoding for the audio in the first set of one or more spectral bands;

performing inverse quantization for the audio in the first set of one or more spectral bands; and

selectively performing an inverse multi-channel transform between the multiple channels of the audio in the first set of one or more spectral bands;

decoding the audio in a second set of one or more spectral bands higher in frequency than the first set of one or more spectral bands, the audio in the second set of one or more spectral bands having been truncated during encoding to leave the audio in the first set of one or more spectral bands, wherein the decoding the audio in the second set of one or more spectral bands uses a different array of noise parameters for each independently coded channel when the multiple channels are independently coded, wherein the decoding the audio in the second set of one or more spectral bands uses an array of noise parameters for a joint coding channel when the multiple channels are jointly coded, and wherein the decoding the audio in the second set of one or more spectral bands includes;

when the multiple channels are independently coded, for each of the multiple independently coded channels, decoding one or more noise parameters of the different array of noise parameters for the independently coded channel, each of the one or more noise parameters indicating a noise parameter value for a frequency band of one or more of the spectral bands in the second set over a time window of the independently coded channel;

when the multiple channels are jointly coded, decoding one or more noise parameters of the array of noise parameters for the joint coding channel, each of the one or more noise parameters indicating a noise parameter value for a frequency band of one or more of the spectral bands in the second set over a time window of the joint coding channel; and

based at least in part upon the noise parameters, generating patterns that represent the audio in the second set of spectral bands;

combining the patterns that represent the audio in the second set of one or more spectral bands with results of decoding the quantized spectral information for the audio in the first set of one or more spectral bands; and

outputting, from the audio decoder, the reconstructed audio for playback.

19. A computer system comprising a processing unit, memory and storage media storing computer-executable instructions for causing the processing unit to perform a method of audio decoding, the method comprising:

receiving encoded audio information for audio in multiple channels;

decoding the encoded audio information to produce reconstructed audio, including;

decoding the audio in a first set of one or more spectral bands, the audio in the first set of one or more spectral bands having been encoded as quantized spectral information, wherein the decoding the audio in the first set of one or more spectral bands includes;

performing entropy decoding for the audio in the first set of one or more spectral bands;

performing inverse quantization for the audio in the first set of one or more spectral bands; and

selectively performing an inverse multi-channel transform between the multiple channels of the audio in the first set of one or more spectral bands;

decoding the audio in a second set of one or more spectral bands higher in frequency than the first set of one or more spectral bands, the audio in the second set of one or more spectral bands having been truncated during encoding to leave the audio in the first set of one or more spectral bands, wherein the decoding the audio in the second set of one or more spectral bands uses a different array of noise parameters for each independently coded channel when the multiple channels are independently coded, wherein the decoding the audio in the second set of one or more spectral bands uses an array of noise parameters for a joint coding channel when the multiple channels are jointly coded, and wherein the decoding the audio in the second set of one or more spectral bands includes;

when the multiple channels are independently coded, for each of the multiple independently coded channels, decoding one or more noise parameters of the different array of noise parameters for the independently coded channel, each of the one or more noise parameters indicating a noise parameter value for a frequency band of one or more of the spectral bands in the second set over a time window of the independently coded channel;

when the multiple channels are jointly coded, decoding one or more noise parameters of the array of noise parameters for the joint coding channel, each of the one or more noise parameters indicating a noise parameter value for a frequency band of one or more of the spectral bands in the second set over a time window of the joint coding channel; and

based at least in part upon the noise parameters, generating patterns that represent the audio in the second set of spectral bands;

combining the patterns that represent the audio in the second set of one or more spectral bands with results of decoding the quantized spectral information for the audio in the first set of one or more spectral bands; and

outputting, from the audio decoder, the reconstructed audio for playback.