Variable Order Short-Term Predictor for MPEG-4 ALS
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
MPEG-4 Audio Lossless coding (ALS) hardware implementation faces high complexity due to wide ranges of short-term predictor filter orders, particularly in the calculation time mismatch between linear prediction coefficients and FIR filter operations, leading to inefficient pipeline schemes and reduced compression ratios.
Innovation Solution
A variable order short-term predictor is implemented using a pipeline stall based on autocorrelation processing time and the radix-2 algorithm, with a pre-decision module determining modified iterations and a loop controller managing the FIR filter operations to optimize processing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the filter order is increased to support wide range of MPEG-4 ALS, then the prediction accuracy is improved, but the device complexity increases
Solution Approach 1:
The patent divides the high-order filter into multiple stages of lower-order filters. Specifically, a 1024-tap filter is segmented into multiple cascaded stages, each handling a portion of the total filter order. This segmentation reduces the complexity of implementing a single high-order filter while maintaining the overall prediction accuracy required by MPEG-4 ALS standard.
Solution Approach 2:
The patent transforms the time-domain high-order filtering operation into frequency-domain processing using Fast Fourier Transform (FFT). By converting the filter operation to the frequency domain, the complexity of implementing a 1024-tap filter in time domain is reduced, as frequency-domain multiplication is computationally more efficient than time-domain convolution for high-order filters.
2Manufacturing precision
If the calculation time of linear prediction coefficients is increased to improve precision, then the manufacturing precision is improved, but the productivity decreases
Solution Approach 1:
The patent pre-calculates and stores the linear prediction coefficients in a lookup table (LUT) during a initialization phase or offline processing. During real-time encoding, these pre-computed coefficients are directly retrieved from the LUT without performing time-consuming autocorrelation calculations, thus maintaining coefficient precision while significantly improving processing speed and productivity.
Solution Approach 2:
The patent implements a dynamic coefficient selection mechanism that adapts the filter order and coefficient precision based on the characteristics of the input audio signal. For signals requiring high precision, the system uses higher-order filters with more precise coefficients, while for simpler signals, lower-order filters suffice. This dynamic adaptation maintains manufacturing precision where needed while improving overall productivity by avoiding unnecessary high-precision calculations.
3Measurement precision
If the filter order is increased to improve prediction accuracy, then the measurement precision is improved, but the loss of time increases
Solution Approach 1:
The patent pre-computes and stores the results of high-order filter operations in lookup tables during an initialization phase. During actual audio encoding, the system retrieves pre-computed filter results from these tables rather than performing time-consuming high-order filtering operations in real-time, thus maintaining prediction accuracy while significantly reducing processing time and avoiding time loss.
Solution Approach 2:
The patent transforms the time-domain high-order filtering into frequency-domain processing using FFT, where the computationally intensive convolution operation becomes a simple multiplication in the frequency domain. This dimensional transformation dramatically reduces the processing time required for high-order filtering while preserving the accuracy of the prediction, thereby eliminating the time loss that would otherwise result from direct time-domain computation.
Data Source
AI summary
The present invention provides a new recursive FIR filter scheme which supports a variable order short-term predictor, and uses a pipeline stall based on the radix-2 algorithm and an autocorrelation processing time for reducing the complexity of MPEG-4 ALS hardware implementation.