Acoustic Echo Canceller Near-End Signal Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Acoustic Echo Canceller systems face delays in detecting the onset of near-end signals, leading to divergent adaptive filter coefficients and ineffective echo cancellation, which disrupts conversation clarity.
Innovation Solution
A method and system that utilize an adaptive filter, error signal, and onset indicator parameters to quickly detect the onset of near-end signals, incorporating the cosine value of the angle between microphone output and acoustic echo estimates, to prevent adaptive filter divergence and enhance Echo Return Loss Enhancement (ERLE) performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the Angle Based Method (ABM) is used to detect the onset of near-end signal, then the detection mechanism is simple, but the detection delay increases causing adaptive filter coefficients to diverge
Solution Approach 1:
The patent changes the detection parameter from angle-based metrics to energy-based metrics (error signal energy and acoustic echo energy). This parameter change enables faster detection response while maintaining computational simplicity, resolving the contradiction between simple detection mechanism and detection delay.
Solution Approach 2:
The patent replaces the angle-based detection mechanism with an energy-based detection mechanism. This substitution fundamentally changes the detection approach from geometric relationships to energy comparisons, achieving faster detection without increasing system complexity.
2Measurement precision
If adaptive filter coefficients are updated continuously, then the echo estimation is accurate, but the coefficients diverge when near-end signal is present
Solution Approach 1:
The patent introduces feedback through the onset detection mechanism that monitors energy levels and provides control signals to the adaptive filter. When near-end signal onset is detected through energy comparison, the feedback stops coefficient updates, preventing divergence while maintaining accuracy during echo-only periods.
Solution Approach 2:
The patent makes the adaptive filter operation dynamic by conditionally enabling or disabling coefficient updates based on near-end signal detection. The filter transitions between active updating mode (when only echo is present) and frozen mode (when near-end signal is detected), maintaining both accuracy and stability.
3Device complexity
If the Near-end Signal Detector (NSD) uses traditional detection methods, then the system complexity is low, but the conversation clarity deteriorates due to delayed detection
Solution Approach 1:
The patent changes the detection parameters from angle-based metrics to energy-based metrics (comparing error signal energy with acoustic echo energy). This parameter change achieves faster detection response that maintains conversation clarity while keeping the NSD system complexity low.
Data Source
AI summary
A method, a system and a computer program product for fast detection of the onset of a near-end signal is provided. An Acoustic Echo Canceller (AEC) attenuates an acoustic echo present in a tele-communication network. The AEC includes an adaptive filter that estimates the acoustic echo and generates an error signal. The error signal is the difference between the acoustic echo and the estimate of acoustic echo plus a near-end signal, if present. The method comprises computing an onset indicator parameter from the error signal and the estimate of acoustic echo. Several other parameters are subsequently calculated by using the onset indicator parameter, the error signal and the estimate of acoustic echo.