Echo Cancellation Using NLMS and Auxiliary Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing echo cancellation methods using normalized least mean square adaptive filtering (NLMS) algorithms often leave residual echo signals due to inaccuracies in estimating far-end signals, affecting call quality.
Innovation Solution
An echo cancellation method that detects the presence of a speech signal in near-end signals and processes far-end signals using the NLMS algorithm to obtain an estimated signal, then subtracts this from the near-end signal to get a residual signal. If the residual signal does not meet a preset output condition, it is attenuated by a first attenuation factor to reduce strong echo signals, and the method also detects the presence of echo signals in near-end signals to avoid unnecessary processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the NLMS algorithm is used to estimate the far-end signal, then the echo cancellation effect is improved, but residual echo signals remain due to estimation errors, affecting call quality
Solution Approach 1:
The patent introduces an auxiliary detector as an intermediary component that works alongside the NLMS algorithm. This detector monitors the residual signal and identifies when echo signals are still present, enabling additional processing only when necessary. The intermediary detector bridges the gap between the NLMS estimation and complete echo elimination, addressing the precision limitation without requiring complete redesign of the estimation algorithm.
Solution Approach 2:
The patent dynamically adjusts processing parameters based on the detected signal conditions. When the auxiliary detector identifies residual echo, the system changes the processing state to apply additional echo cancellation. This parameter change approach allows the system to adapt to the limitations of the NLMS algorithm by activating supplementary processing only when estimation accuracy is insufficient, thereby improving overall echo cancellation without continuously operating at full complexity.
2Reliability
If the NLMS algorithm continuously processes far-end signals, then echo cancellation is maintained, but processor resources are wasted when no echo signal is present
Solution Approach 1:
The patent implements a dynamic processing architecture where the NLMS algorithm's operation state changes based on real-time detection. The auxiliary detector continuously monitors the near-end signal to determine whether echo signals are present. When no echo is detected, the system dynamically switches to a lower-power state, preventing unnecessary NLMS processing. This dynamic adaptation resolves the contradiction by maintaining echo cancellation reliability only when actually needed, thereby conserving processor resources during normal speech transmission without echo.
Solution Approach 2:
The patent extracts the echo detection function into a separate auxiliary detector module that operates independently from the main NLMS processing chain. This extraction allows the system to identify echo presence and selectively activate the computationally intensive NLMS algorithm only when echo signals are detected. By separating the detection function from the processing function, the system avoids continuous high-resource consumption while maintaining reliable echo cancellation when required.
3Device complexity
If the residual signal is always output, then processing simplicity is maintained, but strong residual echo signals degrade call quality
Solution Approach 1:
The patent introduces an auxiliary detector as an intermediary between the residual signal generation and output stages. This detector analyzes the residual signal to determine whether strong echo components remain. Based on this detection, the system selectively applies additional processing only when harmful residual echo is identified. The intermediary detector thus prevents direct output of degraded signals while avoiding unnecessary processing when the residual signal is already clean, resolving the contradiction between simplicity and quality.
Solution Approach 2:
The patent dynamically changes the output signal processing parameters based on the detected residual echo characteristics. When the auxiliary detector identifies strong residual echo, the system adjusts processing parameters to apply additional attenuation or filtering. When residual echo is minimal, the system maintains simple direct output. This conditional parameter adjustment resolves the contradiction by applying complexity only when harmful factors are present, thereby maintaining call quality without permanently increasing device complexity.
Data Source
AI summary
The present disclosure discloses an echo cancellation method and apparatus. The method includes: obtaining a far-end signal and a near-end signal; detecting whether a speech signal exists in the near-end signal; processing the far-end signal according to a normalized least mean square adaptive filtering (NLMS) algorithm to obtain an estimated signal, and subtracting the estimated signal from the near-end signal to obtain a residual signal; after it is detected that the speech signal exists in the near-end signal, detecting whether the residual signal meets a preset output condition; and when it is detected that the residual signal does not meet the preset output condition, multiplying the residual signal by a first attenuation factor to obtain an output signal.