Microphone Beamforming Post-Processing for Out-of-Beam Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video conferencing systems struggle to suppress out-of-beam noise sources due to in-beam acoustic reflections, which are not effectively filtered by spatial filters, leading to disruption during calls.
Innovation Solution
A method involving the derivation of in-beam components from multiple microphone signals, computation of reference and in-beam levels, and application of a post-processing gain to suppress out-of-beam noise while enhancing in-beam audio clarity using beam-forming and post-processing techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If a spatial filter (beam-former) is used to filter out acoustic signals from out-of-beam directions, then out-of-beam noise sources are suppressed, but in-beam acoustic reflections from out-of-beam sources are not filtered and are transmitted un-attenuated
Solution Approach 1:
The patent segments the audio signal processing into multiple stages: first spatial filtering to separate in-beam and out-of-beam components, then individual processing of each component. The in-beam signal path and out-of-beam signal path are separated and processed independently, allowing different filtering strategies to be applied to each segment, thereby resolving the contradiction between suppressing out-of-beam noise and maintaining in-beam signal quality
Solution Approach 2:
The patent introduces an intermediary processing stage that takes the output of the spatial filter and further processes the in-beam component before final combination. This intermediary stage applies additional filtering and gain control to the in-beam signal, ensuring that any remaining out-of-beam reflections are attenuated while preserving the desired in-beam audio, thus improving overall noise filtering effectiveness
2Object-affected harmful factors
If deep-learning based models are applied to filter out-of-beam noise, then noise suppression is improved, but system complexity and computational requirements increase
Solution Approach 1:
The patent replaces complex deep-learning based models with traditional signal processing techniques including spatial filtering, spectral subtraction, and gain control. These conventional methods achieve effective noise suppression without the computational burden and system complexity of deep-learning models, while maintaining acceptable performance for video conferencing applications
Solution Approach 2:
The patent adjusts processing parameters such as beam-width, filter coefficients, and gain values to optimize noise suppression performance. By carefully tuning these parameters, the system achieves effective out-of-beam noise rejection using simpler processing methods, avoiding the need for computationally intensive deep-learning models
Data Source
AI summary
A computer-implemented method of processing an audio signal. The method comprises: receiving from two or more microphones, respective audio signals; deriving a plurality of time-frequency signals from the received audio signals, indexed by frequency, and for each of the time-frequency signals: determining in-beam components of the audio signals; and performing post-processing of the received audio signals, the post-processing comprising: computing a reference level based on the audio signals; computing an in-beam level based on the determined in-beam components of the audio-signals; computing a post-processing gain to be applied to the in-beam components from the reference level and in-beam level; and applying the post-processing gain to the in-beam components.