Partially Adaptive Audio Beamforming for Echo Cancellation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio beamforming systems struggle with effective acoustic echo cancellation when a loudspeaker is in close proximity to a microphone array, leading to suboptimal performance and distortion of desired sound signals due to high echo-to-signal ratios and computational resource intensity.
Innovation Solution
The implementation of a frequency domain beamforming technique with a stored beamforming parameter, such as an inverse covariance matrix, that generates a beamformed audio signal to attenuate undesired sounds from the loudspeaker while minimizing distortion of desired sounds, and updates coefficients based on voice activity to reduce computational and memory resource usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If a typical acoustic echo cancellation system is used to remove echo from microphone array signals, then linear echo portions can be attenuated by up to 20 dB, but the system performs poorly and has suboptimal performance when the loudspeaker is in close proximity to the microphone array due to high echo-to-signal ratios
Solution Approach 1:
The patent segments the echo cancellation task into two parts: a beamforming stage that handles the high echo-to-signal ratio environment by spatially filtering signals from multiple microphones, and a subsequent AEC stage that processes the beamformed output. This segmentation allows each stage to operate in its optimal performance range, with the beamformer reducing the echo level before the AEC processes the remaining echo.
Solution Approach 2:
The beamforming operation is performed as a preliminary action before acoustic echo cancellation. The beamformer pre-processes the microphone array signals to create a beamformed audio signal with reduced echo content, which then serves as the input to the AEC system. This preliminary spatial filtering improves the conditions for subsequent echo cancellation.
2Object-affected harmful factors
If existing beamforming techniques are used to attenuate echo signals, then certain portions of echo can be reduced, but the techniques are computationally and memory resource intensive and difficult to implement in certain audio devices
Solution Approach 1:
The patent applies partial beamforming action by using a simplified beamforming approach that processes only the necessary components of the microphone signals. Rather than implementing full adaptive beamforming algorithms, the system uses a practical beamformer that provides sufficient echo reduction while maintaining computational efficiency for deployment in resource-constrained audio devices.
Data Source
AI summary
Partially adaptive audio beamforming systems and methods are provided that enable improved acoustic echo cancellation of sound played on a loudspeaker that is in close proximity to a microphone array in an audio device. A stored beamformer parameter, such as an inverse covariance matrix, can be utilized by a frequency domain beamformer to generate a beamformed signal. The overall performance and resource usage by the audio device can be optimized.