Full-Duplex Echo Cancellation via Dual-Path Beamforming
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In full-duplex audio communications, the close proximity of microphones and loudspeakers in IoT and room-based applications leads to high echo-to-near-end speech ratios, causing poor audio performance, especially double talk, due to the high acoustic signal propagation from the far-end endpoint being captured by the microphone as echo.
Innovation Solution
Implementing a dual-path processing system where far-field and near-field beam signals are concurrently processed to detect and cancel echo, with adaptive dereverberation and noise reduction techniques, and switching between far-field and near-field output signals based on echo thresholds to optimize speech quality and reduce echo-to-near-end-speech ratios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the microphone and loudspeaker are integrated into one unit with close proximity, then the device structure is simplified and far-field voice interface is achieved, but the echo-to-near-end speech ratio increases causing poor full-duplex performance
Solution Approach 1:
The patent segments the beamforming processing into two independent paths: far-field beamforming for capturing distant speech and near-field beamforming for suppressing acoustic echo. This segmentation allows each path to be optimized for its specific function without interfering with the other, resolving the contradiction between integrated structure and full-duplex performance
Solution Approach 2:
The patent introduces an intermediary switching mechanism that selects between far-field and near-field output signals based on echo detection. This intermediary component mediates between the conflicting requirements of far-field speech capture and echo suppression, enabling the system to adaptively choose the optimal path
2Object-affected harmful factors
If beamforming is optimized to suppress acoustic echo, then echo is reduced, but a degree of freedom is consumed compromising far-field speech pickup, noise reduction, and dereverberation performance
Solution Approach 1:
The patent creates separate near-field and far-field beamforming paths with independent processing chains. The near-field path specifically targets echo suppression while the far-field path maintains optimal speech pickup, noise reduction, and dereverberation capabilities. This segmentation eliminates the trade-off by allowing each path to specialize
Solution Approach 2:
The patent implements dynamic switching between near-field and far-field output signals based on real-time echo detection. When echo is detected above a threshold, the system dynamically switches to near-field output; otherwise, it uses far-field output. This dynamic adaptation allows the system to maintain far-field speech pickup performance while suppressing echo when necessary
Data Source
AI summary
A loudspeaker is driven with a loudspeaker signal to generate sound, and sound is converted to one or more microphone signals with one or more microphones. The microphone signals are concurrently transformed into far-field beam signals and near-field beam signals. The far-field beam signals and the near-field beam signals are concurrently processed to produce one or more far-field output signals and one or more near-field output signals, respectively. Echo is detected and canceled in the far-field beam signals and in the near-field beam signals. When the echo is not detected above a threshold, the one or more far-field output signals are outputted. When the echo is detected above the threshold, the one or more near-field output signals are outputted. A signal based on the one or more output signals is transmitted.