Audio Decoding Apparatus Reducing Computational Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing MPEG Surround audio decoding systems require significant computational resources and power consumption due to the repeated use of orthogonal transforms like MDCT and QMF, which increases the overall computation needed for decoding.
Innovation Solution
The proposed audio decoding apparatus exploits the symmetry of basis functions in the butterfly inverse modified discrete cosine and sine transforms (IMDCT and IMDST) to reduce the computation required for transforming MDCT coefficients into QMF coefficients, specifically by utilizing the symmetry to perform computations for only half of the transform matrix, thereby reducing the computational burden.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If orthogonal transforms (MDCT and QMF) are performed repeatedly to decode audio signals in the MPEG Surround System, then the decoding accuracy and audio quality are improved, but the computational load and power consumption increase significantly
Solution Approach 1:
The patent segments the transform matrix computation into two distinct stages: (1) computing only the first half of the QMF coefficients using MDCT coefficients, and (2) deriving the second half of the QMF coefficients from the first half using symmetry properties. This segmentation reduces the computational load by eliminating redundant calculations while preserving decoding accuracy.
Solution Approach 2:
The patent applies partial action by computing only 50% of the transform matrix (the first half of QMF coefficients) explicitly, and obtaining the remaining 50% (the second half) through symmetry relationships. This partial computation approach reduces power consumption by half for the transform operation while maintaining complete decoding functionality.
2Measurement precision
If the full transform matrix is computed to transform MDCT coefficients into QMF coefficients, then the transformation accuracy is maintained, but the computational complexity increases
Solution Approach 1:
The patent exploits the asymmetric computational approach where the first half of the transform matrix is computed explicitly while the second half is derived through symmetry properties. This asymmetric computation strategy reduces computational complexity from O(N²) to approximately O(N²/2) while maintaining transformation accuracy through the use of valid symmetry relationships.
3Reliability
If complete QMF coefficient computation is performed for residual signal transformation, then the audio reconstruction quality is improved, but the processing time increases
Solution Approach 1:
The patent performs preliminary computation of only the first half of the QMF coefficients, then uses symmetry properties to rapidly derive the second half. This preliminary action approach reduces processing time by eliminating the need to compute the entire transform matrix, while the symmetry-based derivation ensures audio reconstruction quality is maintained.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An orthogonal transform apparatus computes either one of the real and imaginary components of the quadrature mirror filter coefficient contained in a first subinterval of a plurality of subintervals among which a coefficient sequence containing a plurality of quadrature mirror filter coefficients is divided so that the values of basis functions are symmetrically placed, by computing a sum of products of the plurality of modified discrete cosine transform coefficients and the basis functions corresponding to the subinterval, computes the other one of the real and imaginary components of the quadrature mirror filter coefficient contained in the first subinterval and the real and imaginary components of the quadrature mirror filter coefficient contained in another subintervals by performing a butterfly operation using a computed value produced as a result of the sum of products and computes each quadrature mirror filter coefficient by combining the real component and imaginary component thereof.