Dual Sparse Echo Cancellation Filter for Path Change Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional echo cancellers fail to detect echo path changes timely and efficiently, leading to residual echo issues due to high complexity and memory requirements in existing systems.
Innovation Solution
The use of a dual-filter approach with a foreground and background SPARSE adaptive filter, where the foreground filter operates in open- and closed-loop modes, and the background filter adapts aggressively in the presence of double talk, allowing for quick detection and adjustment of echo path changes with reduced memory and complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If a non-SPARSE filter is used to detect echo path changes, then detection timeliness is improved, but memory consumption and system complexity increase enormously
Solution Approach 1:
The patent divides the single filter into two separate SPARSE filters: a foreground filter for real-time echo cancellation and a background filter for detecting echo path changes. This segmentation allows each filter to specialize in its function while maintaining low computational complexity and memory usage, resolving the contradiction between timely detection and system simplicity.
Solution Approach 2:
The background filter performs preliminary detection of echo path changes by monitoring the error signal independently. When the background filter detects a significant echo path change, it triggers an update to the foreground filter. This preliminary action enables timely detection without requiring the main foreground filter to continuously analyze all parameters, thus reducing overall system complexity.
2Measurement precision
If a full-length window is applied to cover the entire echo path delay, then echo cancellation accuracy is improved, but memory space requirements become enormous
Solution Approach 1:
The patent applies SPARSE filtering which concentrates filter energy only in relevant time regions rather than distributing it uniformly across the entire echo path delay. This local quality approach maintains high echo cancellation accuracy by focusing computational resources on the most significant echo components while dramatically reducing the memory space required for filter coefficients.
Solution Approach 2:
The SPARSE filter uses a partial window approach, applying filtering only to the most significant portions of the echo path rather than the entire duration. This partial action maintains adequate echo cancellation performance by addressing the dominant echo components while avoiding the excessive memory requirements of processing the complete echo tail.
3Stability of the object's composition
If the foreground adaptive filter operates in closed-loop mode continuously, then echo cancellation stability is improved, but adaptability to echo path changes deteriorates
Solution Approach 1:
The background filter continuously monitors the error signal and provides feedback about echo path conditions. When the background filter detects significant changes, it triggers updates to the foreground filter's bulk delay parameter. This feedback mechanism enables the stable foreground filter to adapt to echo path changes without compromising its cancellation performance or requiring continuous aggressive adaptation.
Solution Approach 2:
The system dynamically switches between stable operation (foreground filter in closed-loop mode) and adaptation modes (when background filter detects changes). This dynamic behavior allows the system to maintain stability during normal operation while quickly adapting to echo path changes when necessary, resolving the contradiction between stability and adaptability.
Data Source
AI summary
There is provided a method for use by an echo canceller to detect an echo path change and adjust to the echo path change. The method comprises determining a first bulk delay using a SPARSE foreground adaptive filter; configuring the foreground adaptive filter to an open-loop mode; canceling the echo signal based on the first bulk delay using the foreground adaptive filter; determining a second bulk delay of the echo signal using a SPARSE background adaptive filter; configuring the foreground adaptive filter to a closed-loop mode and continuing to cancel the echo signal based on the first bulk delay; configuring the background adaptive filter to the open-loop mode; measuring echo cancellation performance of the foreground adaptive filter and the background adaptive filter; and changing parameters of the foreground adaptive filter if the echo cancellation performance of the background adaptive filter is better than the foreground adaptive filter.