Audio Filter Design Using Symmetric Gain-Surface Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing beamforming techniques struggle with robustness to model mismatch and require rigorous searching to achieve desired output, leading to interference from background noise and unintended signals in audio signals.
Innovation Solution
A computer-implemented method that generates an asymmetrical white noise gain surface, converts it to a symmetrical surface using a whitening function, and transforms extremum points to filter audio signals, using non-binary values to enhance desired speech signals while reducing interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional diagonal loading techniques are applied to beamforming filter, then filter robustness to model mismatch is improved, but rigorous searching is required to achieve desired output
Solution Approach 1:
The patent transforms the asymmetrical white noise gain surface to a symmetrical one through parameter transformation, enabling the use of efficient extremum seeking algorithms. This changes the mathematical parameters of the optimization problem from an asymmetrical constrained surface to a symmetrical one, reducing computational complexity and searching time while maintaining filter robustness through the extremum point identification
Solution Approach 2:
The patent replaces traditional iterative beamforming optimization methods with a mathematical transformation approach. By converting the asymmetrical gain surface to symmetrical form and using closed-form extremum identification, it substitutes complex iterative mechanical searching with more efficient mathematical computation
2Measurement precision
If beamforming techniques are used to enhance desired speech signals, then spatial diversity is utilized, but background noise and unintended signals interfere with the desired speech signals
Solution Approach 1:
The patent introduces asymmetrical white noise gain surface that treats desired speech signals and interference signals differently. By applying asymmetrical weighting and constraint design, it creates different gain characteristics for target signals versus interference signals, enhancing speech while attenuating background noise and unintended signals
Solution Approach 2:
The patent applies local quality optimization by identifying extremum points on the white noise gain surface that correspond to specific spatial locations. The filter design creates locally optimized gain characteristics for different spatial regions, enhancing signals from desired directions while suppressing signals from other directions
Data Source
AI summary
A computer-implemented method executed by data processing hardware causes the data processing hardware to perform operations that include receiving multiple audio signals from a sensor array. The multiple audio signals include a target audio signal and interference audio signals. The data processing hardware then identifies a design constraint based on the multiple audio signals. The desired constraint includes a pass constraint corresponding to the target audio signal and a null constraint corresponding to the interference audio signals. The data processing hardware then compares a design filter weight of the design constraint with a filter weight maximum, designs an audio filter using the desired constraint, and filters the multiple audio signals using the designed audio filter.