SBR Frame Class for Low-Delay Transient Audio Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current audio coding techniques using SBR (Spectral Band Replication) face challenges in managing transients, leading to audible artifacts like pre-echoes, especially at low bit rates, due to the need for balancing temporal and spectral resolution with bit rate constraints.
Innovation Solution
Introducing a new SBR frame class, LD_TRAN, where frame boundaries are synchronized with raster boundaries, and using transient position information to define short grid areas around transients, reducing the need for variable SBR frames and minimizing delay.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the time grid for encoding the spectral envelope is made coarse to reduce bit consumption, then bit rate is reduced, but audible artifacts like pre-echo appear during transients
Solution Approach 1:
The patent applies dynamics by making the time grid adaptive rather than fixed. The grid density automatically increases around detected transient events and decreases in stationary regions. This allows the system to maintain high quality during transients (narrow grid) while saving bits during normal playback (coarse grid), resolving the contradiction between bit rate and artifact prevention.
Solution Approach 2:
The patent implements local quality by applying different grid resolutions to different time regions. Instead of using a uniformly coarse grid throughout the entire signal, the system uses fine-grained resolution locally around transients and coarse resolution elsewhere. This targeted approach prevents artifacts only where needed while maintaining overall bit efficiency.
2Object-affected harmful factors
If the raster area around transients is narrowed to improve transient-to-background ratio, then artifact quality improves, but more bits are required to encode the spectral envelope
Solution Approach 1:
The patent uses dynamic grid adjustment where the raster area width is not fixed but adapts based on transient detection. When a transient is detected, the grid automatically narrows around that specific region. This dynamic behavior allows narrow raster areas only where transients occur (improving quality) while maintaining coarse grids elsewhere (reducing bit consumption).
Solution Approach 2:
The patent segments the time signal into different regions based on transient detection. Each segment receives appropriate grid resolution treatment - fine segmentation around transients and coarse segmentation in between. This segmentation strategy applies narrow raster areas selectively rather than uniformly, improving transient quality without proportionally increasing overall bit consumption.
3Object-affected harmful factors
If variable SBR frames with shifted boundaries are used to handle transients, then transient handling improves, but decoding delay increases due to overlap compensation requirements
Solution Approach 1:
The patent applies dynamics by using fixed frame boundaries combined with dynamic transient detection and localized grid adjustment within those frames. Instead of shifting frame boundaries to follow transients (which causes delay), the system keeps boundaries fixed and adapts the internal grid structure dynamically. This eliminates the need for overlap compensation while still handling transients effectively.
Solution Approach 2:
The patent inverts the conventional approach by not shifting frame boundaries to accommodate transients. Instead, it keeps boundaries fixed and shifts the grid structure within frames to accommodate transients. This inversion eliminates the causality problem and decoding delay associated with variable frames while maintaining effective transient handling.
Data Source
Figure 1
Figure 2~3
Figure 4a~4b
AI summary
The problem of transients can be addressed satisfactorily and a further decoding delay can be reduced for this if use is made of a new SBR frame class in which the frame limits (902a, 902b) are not shifted, that is to say the grid limits are still synchronized with the frame limits (902a, 902b), but in which a statement of the transient position (T) is also used as a syntax element in order to be used, within the frames of this new frame class, to determine the grid limits within these frames at the coder and/or decoder end.