Audio encoder and decoder with long term prediction

• US 8,494,863 B2
• Filed: 12/30/2008
• Issued: 07/23/2013
• Est. Priority Date: 01/04/2008
• Status: Active Grant

• Current Assignee: Dolby International AB (Dolby Laboratories Incorporated)
• Original Assignee: Dolby Laboratories Licensing Corporation (Dolby Laboratories Incorporated)
• Inventors: Purnhagen, Heiko, Kjoerling, Kristofer, Resch, Barbara, Villemoes, Lars, Hedelin, Per, Biswas, Arijit
• Primary Examiner(s): Shah, Paras D
• Application Number:
  US12/811,419
• Publication Number:
  US 20100286990A1
• Time in Patent Office: 1,666 Days
• Field of Search: 704/205, 704/207, 704/219, 704/220, 704/222, 704/223, 704500-504
• US Class Current: 704/500
• 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 19/0212 using orthogonal transformation
• CPC Class Codes: G10L 19/032 Quantisation or dequantisation of spectral components
• CPC Class Codes: G10L 19/035 Scalar quantisation
• CPC Class Codes: G10L 19/26 Pre-filtering or post-filtering

Abstract

The present invention teaches a new audio coding system that can code both general audio and speech signals well at low bit rates. A proposed audio coding system comprises a linear prediction unit for filtering an input signal based on an adaptive filter; a transformation unit for transforming a frame of the filtered input signal into a transform domain; a quantization unit for quantizing a transform domain signal; a long term prediction unit for determining an estimation of the frame of the filtered input signal based on a reconstruction of a previous segment of the filtered input signal; and a transform domain signal combination unit for combining, in the transform domain, the long term prediction estimation and the transformed input signal to generate the transform domain signal.

Claims

1. Audio coding system comprising:

a linear prediction unit for filtering an input signal based on an adaptive filter;

a transformation unit for transforming a frame of the filtered input signal into a transform domain;

a long term prediction unit for determining an estimation of the frame of the filtered input signal based on a reconstruction of a previous segment of the filtered input signal; and

a transform domain signal combination unit for combining, in the transform domain, the long term prediction estimation and the transformed input signal to generate a combined transform domain signal,

a quantization unit for quantizing the combined transform domain signal;

wherein the long term prediction unit comprises;

a long term prediction extractor for determining a lag value specifying the reconstructed segment of the filtered signal that best fits the current frame of the filtered input signal; and

a virtual vector generator to generate an extended segment of the reconstructed signal when the lag value is smaller than a frame length of the transformation unit, wherein the virtual vector generator applies an iterative fold-in fold-out procedure to refine the generated segment of the reconstructed signal, and wherein the audio coding system further comprises a processor coupled to one or more of the linear prediction unit, the transformation unit, the long term prediction unit, the transform domain signal combination unit, or the quantization unit.

2. Audio coding system of claim 1, comprising:

an inverse quantization and inverse transformation unit for generating a time domain reconstruction of the frame of the filtered input signal; and

a long term prediction buffer for storing time domain reconstructions of previous frames of the filtered input signal.

3. Audio coding system of claim 1, wherein

the adaptive filter for filtering the input signal is based on a Linear Prediction Coding (LPC) analysis operating on a first frame length and producing a whitened input signal, and

the transformation applied to the frame of the filtered input signal is a Modified Discrete Cosine Transform (MDCT) operating on a variable second frame length.

4. Audio coding system of claim 3, comprising:

a window sequence control unit for determining, for a block of the input signal, the second frame lengths for overlapping MDCT windows by minimizing a coding cost function for the input signal block.

5. Audio coding system of claim 4, wherein the MDCT window lengths are dyadic partitions of the input signal block.

6. Audio coding system of claim 4, wherein the window sequence control unit is configured to consider long term prediction estimations generated by the long term prediction unit for window length candidates when searching for the sequence of MDCT window lengths that minimizes the coding cost function for the input signal block.

7. Audio coding system of claim 4, comprising a window sequence encoder for jointly encoding MDCT window lengths and window shapes in a sequence.

8. Audio coding system of claim 3, comprising a linear prediction interpolation unit to interpolate linear prediction parameters generated on a rate corresponding to the first frame length so as to match frames of the transform domain signal generated on a rate corresponding to the second frame length.

9. Audio coding system of claim 1, comprising a perceptual modeling unit that modifies a characteristic of the adaptive filter by chirping and/or tilting an LPC polynomial generated by the linear prediction unit for an LPC frame.

10. Audio coding system of claim 1, comprising a time warp unit for uniformly aligning a pitch component in the frame of the filtered signal by resampling the filtered input signal according to a time-warp curve, wherein the transformation unit and the long term prediction unit operate on time-warped signals.

11. Audio coding system of claim 1, comprising a highband encoder for encoding a highband component of the input signal, wherein quantization steps used in the quantization unit when quantizing the transform domain signal are different for encoding components of the transform domain signal belonging to the highband than for components belonging to a lowband of the input signal.

12. Audio coding system of claim 1, comprising:

a frequency splitting unit for splitting the input signal into a lowband component and a highband component; and

a highband encoder for encoding the highband component,

wherein the lowband component is input to the linear prediction unit.

13. Audio coding system of claim 12, wherein the boundary between the lowband and the highband is variable and the frequency splitting unit determines the cross-over frequency based on input signal properties and/or encoder bandwidth requirements.

14. Audio coding system of claim 12, comprising a signal representation combination unit for combining different signal representations covering the same frequency range and generating signaling data indicating how the signal representations are combined.

15. Audio coding system of claim 1, wherein the long term prediction unit comprises a spectral band replication unit for introducing energy into high frequency components of the long term prediction estimations.

16. Audio coding system of claim 1, comprising a parametric stereo unit for calculating a parametric stereo representation of left and right input channels.

17. Audio coding system of claim 1, wherein the quantization unit decides, based on input signal characteristics, to encode the transform domain signal with a model-based quantizer or a non-model-based quantizer.

18. Audio coding system of claim 1, comprising a quantization step size control unit for determining the quantization step sizes of components of the transform domain signal based on linear prediction and long term prediction parameters.

19. Audio coding system of claim 1, wherein the long term prediction unit comprises:

a long term prediction gain estimator for estimating a gain value applied to the signal of the selected segment of the filtered signal,

wherein the lag value and the gain value are determined so as to minimize a distortion criterion.

20. Audio coding system of claim 19, wherein the distortion criterion relates to the difference of the long term prediction estimation to the transformed input signal in a perceptual domain, the distortion criterion being minimized by searching the lag value and the gain value in the perceptual domain.

21. Audio coding system of claim 9, wherein the modified linear prediction polynomial generated by the perceptual modeling unit is applied as MDCT-domain equalization gain curve when minimizing a distortion criterion for determining the lag value.

22. Audio coding system of claim 19, wherein the long term prediction unit comprises a transformation unit for transforming the reconstructed signal of the selected segment into the transform domain.

23. Audio coding system of claim 10, wherein the long term prediction unit resamples the reconstructed filtered input signal based on the time-warp curve received from the time warp unit when the transformation unit is operating on time-warped signals.

24. Audio coding system of claim 1, wherein the long term prediction unit comprises a noise vector buffer and/or a pulse vector buffer.

25. Audio coding system of claim 1, comprising a joint coding unit to jointly encode pitch related information.

29. Audio coding system of claim 25, wherein the pitch related information comprises at least one of long term prediction parameters, harmonic prediction parameters and time-warp parameters.

30. Audio coding system of claim 4, wherein the coding function is a simplistic perceptual entropy.

31. Audio coding system of claim 22, wherein the transformation is a type-IV Discrete-Cosine Transformation.

26. Audio decoder comprising:

a de-quantization unit for de-quantizing a frame of an input bitstream;

a long term prediction unit for determining long term prediction estimation of the de-quantized frame;

a transform domain signal combination unit for combining, in the transform domain, the long term prediction estimation and the de-quantized frame to generate a combined transform domain signal;

an inverse transformation unit for inversely transforming the combined transform domain signal; and

a linear prediction unit for filtering the inversely transformed transform domain signal;

wherein the long term prediction unit comprises;

a long term prediction buffer; and

a virtual vector generator to generate an extended segment of a reconstructed signal stored in the long term prediction buffer when a long term prediction lag value is smaller than a length of the frame wherein the virtual vector generator applies an iterative fold-in fold-out procedure to refine the generated segment of the reconstructed signal, and wherein the audio decoder further comprises a processor coupled to one or more of the de-quantization unit, the long term prediction unit, the transform domain signal combination unit, the inverse transformation unit, or the linear prediction unit.

27. Audio decoding method executed by an audio decoding device, comprising the steps:

de-quantizing a frame of an input bitstream;

determining a long term prediction estimation of the de-quantized frame;

when the a lag value is smaller than a length of the frame, generating an extended segment of a reconstructed signal that is stored in term prediction buffer;

refining the extended segment of the reconstructed signal by applying an iterative fold-in fold-out procedure;

combining, in the transform domain, the long term prediction estimation and the de-quantized frame to generate a combined transform domain signal;

inverse transforming the combined transform domain signal;

filtering the inversely transformed transform domain signal; and

outputting a reconstructed audio signal.

28. Computer program stored in a memory device for causing a processor of an audio decoding device to perform the audio decoding method according to claim 27.