Sub-Band Audio Error Concealment Without Discarding Valid Coefficients
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing error concealment techniques in DAB receivers, such as muting, repetition, and left-right substitution, result in the unnecessary discarding of valid data and require additional memory to store frames for error masking, leading to waste and inefficiency in handling bit errors in sub-band coded digital audio signals.
Innovation Solution
A method that analyzes individual sub-band coefficients to selectively suppress errors, using time-domain filtering and weighted combination of original and modified coefficients to minimize subjective audio quality impairment, employing a cross-fader with variable gain elements and filters to detect and mitigate bit errors without discarding valid data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If frame-level error concealment techniques (muting, repetition, substitution) are used, then error masking is achieved, but valid data is unnecessarily discarded and additional memory is required
Solution Approach 1:
The patent segments the audio frame into individual sub-band coefficients and processes them independently. Instead of discarding the entire frame when an error is detected, only the specific corrupted coefficients are identified and concealed, while valid coefficients are preserved and used in the output.
Solution Approach 2:
The patent applies error concealment selectively to individual coefficients based on their local error characteristics. The error detection and concealment processes are applied locally to each coefficient rather than globally to the entire frame, allowing valid data to be preserved while correcting only the erroneous portions.
2Reliability
If frame-level error concealment techniques (muting, repetition, substitution) are used, then error masking is achieved, but additional memory is required to store frames
Solution Approach 1:
The patent extracts only the necessary information (error indicators) from the received coefficients and generates concealed values on-the-fly using filtering and combination operations. This eliminates the need to store entire frames in memory, as only minimal processing buffers are required.
Solution Approach 2:
The system uses the received coefficients themselves as the source for generating error-concealed values through filtering and combination operations, rather than relying on stored backup frames. The coefficients serve their own error concealment needs through self-processing.
3Loss of substance
If individual sub-band coefficients are analyzed locally, then valid data waste is reduced, but computational complexity increases
Solution Approach 1:
The patent uses simple, computationally inexpensive operations such as high-pass filtering, absolute value calculation, and threshold comparison for error detection. These lightweight operations enable local coefficient analysis without requiring complex computational resources.
Solution Approach 2:
The patent transforms coefficients through simple parameter changes (high-pass filtering to detect discontinuities, absolute value calculation for error indication) rather than requiring complex analysis. These parameter transformations are computationally efficient while effectively identifying errors.
4Reliability
If bit errors are detected and concealed, then audio quality is improved, but discontinuities and click artefacts may be introduced
Solution Approach 1:
The patent dynamically adjusts the combination of original and filtered coefficients based on detected error conditions. The cross-fading mechanism continuously adapts the weighting between original and concealed values to smooth transitions and avoid abrupt discontinuities that cause click artefacts.
Solution Approach 2:
The patent introduces a high-pass filtered version of the coefficients as an intermediary element. This intermediary signal serves as a bridge between the original coefficients and the concealed values, enabling smooth cross-fading that prevents abrupt transitions and click artefacts in the output.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
A decoder and method of decoding a sub-band coded digital audio signal. The decoder comprises: an input, for receiving sub-band coefficients for a plurality of sub-bands of the audio signal; an error detection unit (20), adapted to analyze the content of a sequence of coefficients in one of the sub-bands, to derive for each coefficient an indication of whether the coefficient has been corrupted by an error of a predefined type; an error masking unit (30), adapted to generate from the sequence a modified sequence of coefficients for the sub-band, wherein errors of the predefined type are attenuated; a coefficient combination unit (40), adapted to combine the received coefficients and the modified coefficients, in dependence upon the indication of error; and a signal reconstruction unit (50), adapted to reconstruct the audio signal using the combined coefficients.