Binaural Beamformer Optimized with Spatial Information
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hearing assistance systems face performance degradation due to inaccurate estimation of signal correlation matrices and imperfect voice activity detection, leading to suboptimal noise reduction and speech quality in noisy environments.
Innovation Solution
An adaptive binaural beamformer using a multichannel Wiener filter optimized with a priori spatial information, formulated as a quadratically constrained quadratic program, and solved with a low-complexity iterative dual decomposition algorithm, which balances noise reduction and speech quality while being robust to estimation errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional beamforming methods are used, then device complexity is reduced, but noise reduction performance and speech quality deteriorate due to inaccurate signal correlation matrix estimation
Solution Approach 1:
The patent changes the optimization parameters from signal correlation matrix estimation to spatial spectrum estimation using a priori spatial information. This parameter transformation allows the system to achieve better noise reduction performance while avoiding the complexity of accurate signal correlation matrix estimation under imperfect voice activity detection conditions
Solution Approach 2:
The patent performs preliminary estimation of spatial spectrum using a priori spatial information about sound sources before the main beamforming operation. This preliminary action provides robust directional information that guides the optimized beamforming process, reducing dependence on accurate real-time signal correlation matrix estimation
2Reliability
If perfect voice activity detection is implemented, then speech quality improves, but computational complexity and system requirements increase
Solution Approach 1:
The patent introduces spatial spectrum estimation as an intermediary between the microphone signals and the beamforming output. This intermediary uses a priori spatial information to provide robust directional guidance, allowing the system to achieve good speech quality without requiring perfect voice activity detection
Solution Approach 2:
The patent transforms the problem from relying on perfect voice activity detection to relying on spatial spectrum estimation with a priori spatial information. This parameter change reduces the stringency of voice activity detection requirements while maintaining speech quality
3Reliability
If accurate signal correlation matrix estimation is performed, then beamforming performance improves, but computational load and processing time increase
Solution Approach 1:
The patent changes the optimization from accurate signal correlation matrix estimation to spatial spectrum optimization using a priori spatial information. This parameter transformation maintains beamforming performance while reducing computational load for real-time processing
Solution Approach 2:
The patent performs preliminary spatial spectrum estimation using a priori spatial information before the main beamforming computation. This preliminary action provides robust directional constraints that simplify the main optimization process, improving processing speed
4Reliability
If optimized beamforming with a priori spatial information is used, then noise reduction and speech quality improve, but computational complexity increases
Solution Approach 1:
The patent segments the beamforming optimization into two parts: spatial spectrum estimation using a priori spatial information, and then beamforming filter optimization based on this spatial spectrum. This segmentation makes the overall complex problem more manageable and implementable
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A hearing assistance system includes an adaptive binaural beamformer based on a multichannel Wiener filter (MWF) optimized for noise reduction and speech quality criteria using a priori spatial information. In various embodiments, the optimization problem is formulated as a quadratically constrained quadratic program (QCQP) aiming at striking an appropriate balance between these criteria. In various embodiments, the MWF executes a low-complexity iterative dual decomposition algorithm to solve the QCQP formulation.