Beamformer Coordination for Noise Estimation and Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current two-microphone noise suppression processes in mobile phones often fail to accurately estimate speech and noise spectra, especially in transient background noise conditions, leading to reduced speech intelligibility and voice distortion.
Innovation Solution
The system dynamically analyzes microphone signals to generate multiple acoustic beams, selecting pairs that achieve both good voice separation and noise matching, using beamforming and adaptive noise estimation to improve noise suppression accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a two-microphone noise suppression process is used, then the device complexity is reduced, but the measurement precision of noise and speech spectra deteriorates
Solution Approach 1:
The patent divides the microphone array into multiple independent beamformers, each responsible for capturing signals from different spatial directions. This segmentation allows the system to analyze and estimate noise and speech spectra from multiple independent channels, improving measurement precision while maintaining manageable device complexity through modular architecture.
Solution Approach 2:
The patent introduces spatial dimension by using multiple beamformers oriented in different directions. Instead of relying on temporal analysis from a single microphone channel, the system exploits the spatial distribution of sound sources to improve spectrum estimation accuracy, adding a dimensional aspect to the signal processing approach.
2Device complexity
If a two-microphone noise suppression process is used, then the device complexity is reduced, but the reliability of noise estimation in transient noise conditions deteriorates
Solution Approach 1:
By segmenting the signal processing into multiple beamformers, each capturing different spatial characteristics, the system can cross-validate noise estimates across multiple channels. This improves reliability in transient noise conditions where a single microphone might be misled by temporary acoustic variations.
Solution Approach 2:
The patent implements a feedback mechanism where the system continuously monitors the performance of multiple beamformers and dynamically selects or weights their contributions based on current acoustic conditions. This feedback loop enhances the reliability of noise estimation by adapting to transient noise scenarios in real-time.
3Measurement precision
If multiple acoustic beams are generated and analyzed, then the measurement precision of voice separation and noise matching is improved, but the device complexity increases
Solution Approach 1:
The patent segments the complex beam analysis task into simpler sub-tasks, where each beamformer independently processes signals from its specific direction. This division of labor improves voice separation accuracy by dedicating specific processing resources to specific spatial zones, while the modular structure prevents the overall system complexity from becoming unmanageable.
Solution Approach 2:
The patent designs the beamformers to be universal processing units that can handle multiple functions: voice detection, noise estimation, and spectrum analysis. This multi-functionality reduces the need for separate specialized components, thereby improving measurement precision without proportionally increasing device complexity.
4Measurement precision
If multiple acoustic beams are generated and analyzed, then the measurement precision of noise matching is improved, but the device complexity increases
Solution Approach 1:
The patent segments the noise matching process into independent comparisons between beam pairs, where each beamformer's output is analyzed separately for noise characteristics. This segmentation improves noise matching accuracy by enabling detailed per-beam analysis while keeping the coordination complexity manageable through systematic pairwise comparison.
Data Source
AI summary
An audio system has a housing in which are integrated a number of microphones. A programmed processor accesses the microphone signals and produces a number of acoustic pick up beams based groups of microphones, an estimation of voice activity and an estimation of noise characteristics on each beam. Two or more beams including a voice beam that is used to pick up a desired voice and a noise beam that is used to provide information to estimate ambient noise are adaptively selected from among the plurality of beams, based on thresholds for voice separation and thresholds for noise-matching. Other embodiments are also described and claimed.