Spatial Correlation Matrix Estimation Using Weighted Noise Masks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional spatial correlation matrix estimation methods fail to accurately remove background noise from observation signals, leading to inaccurate estimation of target sound source spatial correlation matrices.
Innovation Solution
A spatial correlation matrix estimation device and method that utilize weighting coefficients to separate and remove background noise from observation signals, using first and second masks to calculate weighted feature value matrices, resulting in an accurate estimation of target sound source spatial correlation matrices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the conventional method of subtracting the average noise feature value matrix from the average target sound feature value matrix is used, then the spatial correlation matrix estimation process is simple, but the estimation accuracy deteriorates because the noise effect is not accurately removed
Solution Approach 1:
The invention changes the parameters of the noise removal process by introducing time average operations and weighting coefficients. Instead of directly subtracting noise matrices, the system calculates time averages of masked observation feature value matrices with different weighting coefficients (α for target sound, β for noise) to accurately estimate the spatial correlation matrix while maintaining computational feasibility
Solution Approach 2:
The invention implements a feedback mechanism by iteratively adjusting the weighting coefficients α and β based on the observed signals and masks. The system uses the observed feature value vectors and calculated masks to continuously refine the noise removal process, ensuring that the noise effect is accurately canceled while preserving target sound information
2Measurement precision
If weighting coefficients are introduced to accurately remove noise, then the estimation accuracy improves, but the calculation complexity increases
Solution Approach 1:
The invention segments the noise removal process into distinct computational stages: (1) calculating observation feature value vectors from observation signals, (2) estimating masks for target sound and noise, (3) calculating weighted average matrices with coefficients α and β, and (4) subtracting the weighted noise matrix from the weighted target sound matrix. This segmentation makes the complex process more manageable and implementable
Solution Approach 2:
The invention performs preliminary calculations of the time averages of the masked observation feature value matrices before the final subtraction operation. By pre-calculating the weighted average target sound feature value matrix and the weighted average noise feature value matrix, the system reduces the complexity of the final estimation step while maintaining high accuracy
Data Source
AI summary
An observation feature value vector is calculated based on observation signals recorded at different positions in a situation in which target sound sources and background noise are present in a mixed manner; masks associated with the target sound sources and a mask associated with the background noise are estimated; a spatial correlation matrix of the target sound sources that includes the background noise is calculated based on the masks associated with the observation signals and the target sound sources; a spatial correlation matrix of the background noise is calculated based on the masks associated with the observation signals and the background noise; and a spatial correlation matrix of the target sound sources is estimated based on the matrix obtained by weighting each of the spatial correlation matrices by predetermined coefficients.