Audio Processor Bandwidth Extension Steering Factor
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing bandwidth extension techniques, such as WESPE, can produce unpleasant artifacts like busyness and roughness, especially for noisy signals, due to the challenges in controlling and steering the high-frequency fine structure generation.
Innovation Solution
The proposed solution involves an audio processor that combines two types of excitations: one generated using WESPE and another from randomly generated noise. This mixture is controlled using a steering factor derived from the baseband coder's outputs, particularly from CELP coding parameters, to optimize the mixing process without requiring additional bit-rate.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If WESPE is used to generate high-frequency fine structure, then bandwidth extension is achieved, but unpleasant artifacts like busyness and roughness occur
Solution Approach 1:
The patent combines WESPE-generated excitation with noise-generated excitation to create a mixed high-frequency signal. This merging of two different excitation sources allows the system to benefit from the harmonic structure preservation of WESPE while incorporating the randomization properties of noise to reduce unpleasant artifacts like busyness and roughness.
Solution Approach 2:
The patent introduces a steering factor that dynamically controls the mixing ratio between WESPE excitation and noise excitation. By adjusting this parameter, the system can adaptively change the composition of the high-frequency signal to optimize quality and minimize artifacts under different signal conditions.
2Stability of the object's composition
If non-linearity operations are used to preserve harmonicity, then some consistency is maintained, but the process becomes difficult to control and steer
Solution Approach 1:
The patent introduces a linear mixing process as an intermediary between the excitation generation and the final high-frequency signal. Instead of relying solely on difficult-to-control non-linear operations, the system uses a controllable linear combination of WESPE and noise excitations, making the process easier to steer while maintaining harmonicity through the WESPE component.
Solution Approach 2:
The patent makes the excitation mixing dynamic by using a steering factor that can be adjusted based on signal conditions. This allows the system to adaptively control the balance between harmonicity preservation and artifact reduction, providing flexibility that static non-linear operations cannot achieve.
3Productivity
If copy-up or mirroring of low-frequency content is used, then high-frequency content is generated, but the original fine structure does not fit
Solution Approach 1:
The patent uses WESPE to copy the temporal envelope and pulse positions from the low-frequency signal to generate high-frequency excitation. This copying approach efficiently generates high-frequency content while preserving the harmonic structure, and the subsequent mixing with noise refines the fine structure accuracy to reduce artifacts.
Data Source
Figure 1A~1B
Figure 2
Figure 3
AI summary
An audio processor or coder for processing or coding a signal, the processor or coder comprising: a band-limited entity or coder configured to provide or code a band-limited signal of the signal; and a bandwidth extension entity or coder configured to provide or code an extended-band signal of the signal, the provided or decoded extended-band signal comprising a mixture of a first extended-band excitation and a second extended-band excitation; wherein the bandwidth extension entity or coder is configured to generate the first extended-band excitation and a noise generator configured to generate random noise as the second extended-band excitation; wherein the mixture is controlled via a steering factor derived from a characteristic output or parameter of the band-limited entity or coder.