Echo Cancellers Detecting Path Changes With Short Adaptive Filters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing echo cancellers face difficulties in detecting echo path changes, which can lead to misinterpretation of new echoes as double talk, resulting in suboptimal convergence and perceived echo during handsfree calls.
Innovation Solution
An additional short adaptive filter is introduced to quickly detect echo path changes by distinguishing between echo path changes and double talk through error signal analysis, with the output feeding an Echo Path Change detector that applies Non-Linear Processing to mask increased errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a long adaptive filter is used to model the echo path, then echo cancellation accuracy is improved, but convergence speed decreases when echo path changes occur
Solution Approach 1:
The patent divides the single long adaptive filter into two separate filters: a long filter for accurate echo cancellation and a short filter for rapid echo path change detection. This segmentation allows each filter to be optimized for its specific function, resolving the contradiction between accuracy and speed.
Solution Approach 2:
The short filter uses only a portion of the echo path information (the most recent, most relevant portion) to quickly detect changes. This partial action enables fast convergence without requiring the full computational resources of the long filter, thus improving speed while maintaining sufficient accuracy for change detection.
2Reliability
If the echo canceller re-converges quickly after echo path changes, then echo perception is reduced, but misinterpretation of new echoes as double talk increases
Solution Approach 1:
The short filter acts as an intermediary between the acoustic environment and the long filter. It quickly adapts to echo path changes and provides early warning signals, allowing the system to prepare for re-convergence without the long filter mistakenly interpreting the transition as double talk.
Solution Approach 2:
The short filter performs preliminary detection of echo path changes before the long filter completes its convergence. This preliminary action allows the system to anticipate and prepare for echo path changes, reducing misinterpretation and improving overall reliability.
3Reliability
If adaptive filtering is used to track echo path changes, then echo cancellation performance is maintained, but computational complexity increases
Solution Approach 1:
Instead of using a single long adaptive filter that continuously tracks all echo path details, the patent employs a short filter that tracks only the most recent and relevant portion of the echo path. This partial tracking reduces computational complexity while maintaining sufficient performance for detecting changes.
Solution Approach 2:
The patent segments the adaptive filtering function into two parts: the long filter handles steady-state echo cancellation with lower update frequency, while the short filter handles dynamic change detection with higher update frequency. This segmentation reduces overall computational complexity by distributing the workload appropriately.
Data Source
AI summary
An echo canceller comprising a first adaptive filter having N_long coefficients for converging to an echo path, a non-adaptive filter representing a direct echo path portion captured by the first adaptive filter and having N_short default coefficients, where N_long>N_short, for quick convergence of the echo canceller at start-up, wherein the default coefficients are replaced by the first N_short coefficients from the first adaptive filter responsive to an improvement in echo return loss enhancement (ERLE) of the first adaptive filter, a second adaptive filter having N_short default coefficients for modeling the direct echo path and providing an indication of double-talk and echo path changes, decision logic for receiving error signal outputs from the first and second adaptive filters and the non-adaptive filter and in response distinguishing between echo path changes and double-talk, and a non-linear processor for attenuating signals responsive to input from the decision logic.