Transient-Dependent Window Overlap for Low-Delay Audio Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio encoding/decoding technologies face challenges in minimizing delay and reducing pre-echo noise while maintaining coding quality, especially in the presence of transients, due to sub-optimum window length choices and look-ahead requirements.
Innovation Solution
An apparatus and method that adaptively select window overlap lengths based on transient locations within frames, allowing for a range of overlap widths and transform lengths to minimize pre-echo noise and reduce delay by using a transient location detector and controller to synchronize encoder and decoder.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If long transform is used for stationary signal passages, then coding efficiency is improved, but temporal spread of coding error around transients increases
Solution Approach 1:
The patent applies dynamics by making the transform length adaptive rather than fixed. The system dynamically switches between long transforms (for stationary signals) and short transforms (for transient signals) based on signal characteristics detection, thereby optimizing both coding efficiency and transient handling performance
Solution Approach 2:
The patent applies local quality by using different transform lengths for different portions of the signal. Long transforms are used for stationary portions to maximize coding efficiency, while short transforms are used for transient portions to minimize temporal spread of coding errors, ensuring each region is processed with the most appropriate parameters
2Object-generated harmful factors
If short transform is used for transient signal passages, then temporal spread of coding error is reduced, but coding efficiency decreases
Solution Approach 1:
The system dynamically adjusts transform length based on transient detection. When transients are detected, short transforms are applied to minimize temporal spread of coding errors. During stationary periods, the system switches back to long transforms to maximize coding efficiency, creating an adaptive solution that optimizes both parameters across different signal regions
3Adaptability or versatility
If asymmetric transition windows are used for block switching, then transform length switching is enabled, but encoder look-ahead delay increases
Solution Approach 1:
The patent extracts and removes the asymmetric transition windows from the block switching process. By eliminating the start and stop windows that caused the look-ahead delay requirement, the system enables transform length switching without the associated time penalty, using simple rectangular windows instead
4Object-generated harmful factors
If reduced overlap is used for non-stationary signals, then time extension of transform is limited, but coding quality for stationary signals deteriorates
Solution Approach 1:
The patent applies dynamics by making the overlap width adaptive. The system uses reduced overlap (similar to short transform overlap) when transient signals are detected to limit time extension of the transform. Conversely, full overlap is used for stationary signals to maintain high coding quality, creating an optimal balance between the two competing requirements
Data Source
Figure 1A
Figure 1B
Figure 1C
AI summary
An apparatus for encoding an audio or image signal, comprises: a controllable windower (102) for windowing the audio or image signal to provide the sequence of blocks of windowed samples; a converter (104) for converting the sequence of blocks of windowed samples into a spectral representation comprising a sequence of frames of spectral values; a transient location detector (106) for identifying a location of a transient within a transient look-ahead region of a frame; and a controller (108) for controlling the controllable windower (102) to apply a specific window having a specified overlap length to the audio or image signal in response to an identified location (210-213) of the transient, wherein the controller (108) is configured to select the specific window from a group of at least three windows comprising a first window (201) having a first overlap length (203), a second window (215) having a second overlap length (218), and a third window (224) having a third overlap length (229) or having no overlap, wherein the first overlap length (203) is greater than the second overlap length (218), and wherein the second overlap length (218) is greater than the third overlap length (229) or greater than an overlap of zero, wherein the specific window is selected based on the transient location such that one of two time-adjacent overlapping windows has coefficients at the location of the transient and the other of the two time-adjacent overlapping windows has second window coefficients at the location of the transient, wherein the second coefficients are at least nine times greater than the first coefficients.