Stereo Acoustic Echo Cancellation Using Correlation-Based Filter Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional stereo acoustic echo cancellation (AEC) systems face challenges with the non-uniqueness issue, particularly when microphone signals are highly correlated, leading to slow convergence or failure of adaptive filtering, and distortion-based de-correlation methods can affect voice quality and alter the stereo spatial image.
Innovation Solution
The implementation of a system that uses correlation-based adaptation control for stereo and mono adaptive filters, where the degree of correlation between input signals determines the adaptation rate and operation of the filters, allowing them to operate in parallel and select the output signal with lower power to achieve effective echo cancellation without distorting the input signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional multi-channel adaptive filtering is used for stereo AEC, then echo cancellation is achieved, but convergence becomes slow or fails when signals are highly correlated
Solution Approach 1:
The patent segments the adaptive filtering process into two independent paths: a stereo adaptive filter that processes correlated signal pairs and a mono adaptive filter that processes individual signals. This segmentation allows each filter to operate independently, preventing the convergence failure that occurs in conventional multi-channel filtering when signals are highly correlated. The mono filter provides a fallback convergence path when the stereo filter encounters non-uniqueness issues.
Solution Approach 2:
The patent dynamically changes the adaptation parameters of the filters based on the correlation level between stereo signals. When correlation exceeds a threshold indicating non-uniqueness, the system switches from stereo adaptive filtering to mono adaptive filtering, effectively changing the filter type parameter to maintain convergence. This parameter adaptation resolves the convergence failure by selecting the appropriate filtering mode based on signal conditions.
2Productivity
If stereo de-correlation is used to address non-uniqueness, then filter convergence is improved, but voice quality and stereo spatial image are distorted
Solution Approach 1:
Instead of applying de-correlation to the input signals, the patent segments the processing into parallel stereo and mono filter paths. The mono adaptive filter processes individual channels independently without modifying the original stereo signal structure. This segmentation approach achieves convergence without requiring signal distortion, as the mono filter can converge on single-channel signals even when stereo channels are highly correlated.
Solution Approach 2:
The patent creates a copy of the signal processing path using a mono adaptive filter that mirrors the stereo filter's functionality but operates on individual channels rather than correlated pairs. This copying approach provides an alternative convergence path without needing to modify or distort the original stereo signals, thereby maintaining voice quality and spatial image integrity while achieving reliable convergence.
3Reliability
If de-correlation processing is applied to reduce signal correlation, then adaptive filtering convergence is enabled, but the stereo spatial image is altered
Solution Approach 1:
The patent segments the signal processing into independent stereo and mono filter paths, allowing the system to select the appropriate path based on correlation conditions. The mono filter path processes signals without altering the original stereo spatial structure, as it operates on individual channels rather than modifying the spatial relationship between channels. This segmentation enables convergence while preserving the stereo spatial image.
Solution Approach 2:
Instead of modifying the input signals to reduce correlation (the conventional de-correlation approach), the patent inverts the problem by processing the correlated signals through separate stereo and mono filter paths. The mono filter path serves as an alternative that doesn't require signal modification, thereby achieving convergence without altering the stereo spatial image structure.
Data Source
AI summary
Example apparatus and methods concern performing stereo acoustic echo cancellation using a correlation based filter adaptation control approach and without using stereo de-correlation. An embodiment includes a stereo adaptive filter that produces an echo removed microphone signal from received audio signals. The embodiment includes a mono adaptive filter that produces an echo removed microphone signal from the received audio signals. A correlation detector determines a level of correlation between the received signals and provides a signal to an adaptive filter controller. The adaptive filter controller controls how the stereo adaptive filter and the mono adaptive filter adapt audio echo cancellation as a function of the correlation between the received signals. A signal selector may select for output the signal from either the stereo adaptive filter or the mono adaptive filter based, for example, on the power level of the signals.