Multichannel Audio Upmixing via Orthogonal Matrix Decorrelation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current audio signal processing techniques fail to effectively generate high-quality diffuse sound fields, resulting in uneven amplitude and artificial sound distribution around a listener, especially when upmixing from fewer input channels to multiple output channels.
Innovation Solution
A system of linear mixing equations is employed to generate audio signals for a diffuse sound field, using a matrix with coefficients derived from orthogonal vectors and psychoacoustic decorrelation methods to create a seamless and immersive sound experience across multiple channels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the same audio signal is reproduced through multiple loud speakers to create a sound field, then the setup is simple and easy to implement, but the resulting sound field has widely varying amplitude and seems artificial
Solution Approach 1:
The patent applies dynamics by making the signal processing adaptive rather than static. The system dynamically adjusts the audio signals sent to each loud speaker based on real-time analysis of the sound field characteristics. The signal generator modifies signals to account for variations in acoustic environment, listener position, and loud speaker responses, creating a more natural and consistent diffuse sound field throughout the listening area.
Solution Approach 2:
The patent changes multiple parameters including signal amplitude, phase, frequency content, and temporal characteristics for each loud speaker output. By independently adjusting these parameters, the system transforms identical signals into differentiated signals that collectively create a realistic diffuse sound field. The signal generator applies different processing parameters to each channel to achieve uniform sound distribution and eliminate artificial characteristics.
2Reliability
If audio signals are upmixed from N channels to M channels (M>N), then the immersive sound experience is improved, but the amplitude becomes uneven and artifacts appear
Solution Approach 1:
The patent implements feedback through the sound field analysis component that continuously monitors the acoustic environment and listener position. This analysis information feeds back to the signal generator, which adjusts the upmixed signals in real-time to maintain amplitude uniformity. The feedback loop enables the system to compensate for variations introduced during upmixing and adapt to changing listening conditions, preventing artifacts and ensuring consistent sound distribution across all M channels.
Solution Approach 2:
The patent applies preliminary action by pre-processing the audio signals through the signal generator before they reach the loud speakers. The system performs preliminary adjustments to signal levels, phases, and spectral content based on predicted listening conditions and loud speaker characteristics. This proactive signal modification prevents amplitude unevenness and artifacts from occurring in the first place, rather than attempting to correct them after upmixing.
3Reliability
If decorrelation techniques are applied to create diffuse sound fields, then the artificial sound distribution is reduced, but the processing complexity increases
Solution Approach 1:
The patent applies universality by designing the signal generator to perform multiple functions within a single integrated component. The signal generator simultaneously handles upmixing, decorrelation, amplitude normalization, phase adjustment, and adaptive processing for all M channels. This multi-functional approach achieves natural-sounding diffuse sound fields without requiring separate dedicated devices for each processing task, thereby reducing overall system complexity while maintaining sound field naturalness.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
A system of linear equations is used to upmix a number N of audio signals to generate a larger number M of audio signals that are psychoacoustically decorrelated with respect to one another and that can be used to improve the representation of a diffuse sound field. The linear equations are defined by a matrix that specifies a set of vectors in an M dimensional space that are substantially orthogonal to each other. Methods for deriving the system of linear equations are disclosed.