Two-Channel Noise Removal via PSD Estimation and Adaptive Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for removing noise from two-channel sound signals, such as those using minimum statistics, MCRA, binaural multichannel Wiener filters, or voice activity detectors, are inefficient in separating diffuse noise and interference noise while maintaining directional cues.
Innovation Solution
A method that estimates power spectral density of diffuse noise, removes it from channel signals using weighted gains, and employs an adaptive filter to isolate interference noise, maintaining signal directionality through recursive averaging and NLMS algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multichannel Wiener filters or voice activity detectors are used to remove noise, then noise removal effectiveness is improved, but device complexity and computational operations increase significantly
Solution Approach 1:
The patent segments the noise removal process into distinct stages: first removing diffuse noise using PSD estimation, then removing interference noise using adaptive filters. This segmentation allows each stage to focus on specific noise types with simpler algorithms, avoiding the need for complex multichannel Wiener filters that attempt to handle all noise types simultaneously.
Solution Approach 2:
The patent introduces an intermediary approach by using minimum statistics and PSD estimation as intermediate steps between raw signal reception and final noise removal. These intermediaries provide simplified noise characteristics that guide subsequent filtering operations, reducing overall computational complexity while maintaining effectiveness.
2Reliability
If complex noise removal algorithms are applied, then noise removal performance is improved, but directional information and spatial cues may be lost
Solution Approach 1:
The patent applies different processing strategies to different noise components: diffuse noise is removed using PSD estimation that preserves spatial characteristics, while interference noise is removed using adaptive filters that maintain directional cues. This local quality approach ensures that each processing stage is optimized for its specific noise type without compromising overall directional information.
Solution Approach 2:
The patent uses dynamic adaptive filters that adjust their parameters based on incoming signal characteristics while maintaining directional information. The filters dynamically adapt to changing noise conditions without fixing spatial cues, allowing effective noise removal that responds to real-time signal variations while preserving directional integrity.
3Device complexity
If simple noise removal methods are used, then device complexity is reduced, but the ability to handle multiple interference signals is insufficient
Solution Approach 1:
The patent implements continuous noise removal through recursive PSD estimation and ongoing adaptive filtering. This continuous processing allows the system to handle multiple interference signals over time without requiring complex batch processing algorithms, maintaining simplicity while achieving versatility in handling various noise conditions.
Solution Approach 2:
The patent changes key parameters such as PSD estimation values and adaptive filter coefficients in response to different noise conditions. By dynamically adjusting these parameters, the system can handle multiple types of interference signals using the same basic algorithmic framework, achieving versatility without increasing fundamental device complexity.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
A method of removing noise from a two-channel signal includes receiving channel signals constituting the two-channel signal; obtaining a noise signal for each channel by removing a target signal from each channel signal by subtracting another channel signal multiplied by a weighted value from each channel signal; estimating a power spectral density (PSD) of diffuse noise from each channel signal; obtaining a target signal including an interference signal for each channel by removing the diffuse noise from each channel signal using the estimated PSD of the diffuse noise; obtaining the interference signal for each channel by removing the diffuse noise from the noise signal for each channel using the estimated PSD of the diffuse noise; and removing the interference signal from the target signal including the interference signal for each channel.