Acoustic Echo Cancellation Convergence in Beamforming Audio Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio conferencing systems with beamforming struggle to achieve rapid initial convergence of acoustic echo cancellers across multiple look directions, leading to unstable performance and echo effects due to the need for re-convergence when talkers change positions or new talkers become active, which is not effectively addressed by prior methods.
Innovation Solution
A method that captures loudspeaker signals and feedback during initial system use for concurrent adaptation of the acoustic echo canceller across all echo paths, utilizing available processing resources during periods of low workload, such as double-silence or double-talk, to train the echo canceller in background modes without disrupting real-time operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If acoustic echo cancellation is performed on all microphone signals in parallel, then echo attenuation is improved, but computational complexity increases significantly
Solution Approach 1:
The patent divides the echo cancellation process into two stages: first performing AEC on individual microphone signals, then performing a second AEC on the beamformed output. This segmentation allows the system to handle computational complexity in manageable steps while achieving the required 40 dB echo attenuation through cumulative processing
2Device complexity
If acoustic echo cancellation is performed on the spatially filtered signal at the beamformer output, then computational complexity is reduced, but the transfer function becomes time varying as the beamformer changes look direction
Solution Approach 1:
The patent performs preliminary AEC processing on individual microphone signals before the beamformer processes them. This preliminary action captures echo paths that are relatively stable regardless of beamformer look direction, establishing a foundation that reduces the burden on the second AEC stage and mitigates the time-varying nature of the transfer function
Solution Approach 2:
The patent implements a two-stage adaptive system where the first AEC adapts to individual microphone echo paths and the second AEC adapts to the beamformed output. This dynamic multi-stage adaptation allows the system to handle changing conditions as the beamformer changes look direction while maintaining computational efficiency
3Measurement precision
If the echo canceller is trained for each sector separately, then convergence accuracy is improved, but total training time increases significantly
Solution Approach 1:
The patent combines multiple echo path training processes into a unified two-stage AEC system. By merging the individual microphone AEC processes and the beamformed output AEC into a coordinated system, the patent achieves convergence for all sectors simultaneously rather than sequentially, reducing total training time while maintaining accuracy through the cumulative effect of both stages
4Ease of operation
If the beamformer points to a particular spatial sector during conversation, then spatial directivity is achieved, but the echo canceller must re-converge when talker position changes
Solution Approach 1:
The patent creates a universal echo cancellation system that processes echoes from all directions through the two-stage AEC approach. The first stage handles echoes at the individual microphone level regardless of direction, while the second stage handles the beamformed output, creating a multi-functional system that maintains echo cancellation effectiveness across all spatial sectors without requiring re-convergence
Data Source
AI summary
A method is set forth for reducing the total acoustic echo cancellation convergence time for all look directions in a microphone array based full-duplex system. The method is based on capturing the loudspeaker signal due to the first far-end speech bursts when the conferencing system is first used, as well as the corresponding loudspeaker feedback signals in the individual microphones. The captured signals are then used for consecutive adaptation of the acoustic echo canceller on all echo paths corresponding to all look directions of the beamformer, thereby training the AEC. This training process can be executed concurrently with normal phone operation, for example, as a background process that utilizes available processing cycles.