Microphone Array Source Separation for Same-Direction Interference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current beamforming algorithms struggle to distinguish between desired sound signals and interference signals when both are directly in front of a mobile apparatus, leading to compromised recording and call quality.
Innovation Solution
A sound receiving processing method and apparatus utilizing a microphone array with beamforming, blind signal separation, and sound feature correlations to distinguish sound sources, extract relevant signals, and determine the relative distance of sound sources, allowing for selective retention of desired sound signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If beamforming algorithm is used to eliminate ambient noise, then noise reduction is improved, but interference signals from the same direction cannot be distinguished from desired sound signals
Solution Approach 1:
The patent segments the sound signal processing into multiple stages: beamforming for initial noise reduction, sound feature extraction to identify characteristic features of sound sources, correlation analysis to distinguish between different sound sources based on their feature patterns, and blind source separation to isolate individual sound sources. This segmentation allows the system to handle both ambient noise and interference signals from the same direction differently through multiple processing layers.
Solution Approach 2:
The patent introduces sound feature correlations as an intermediary mechanism between the beamforming output and the final sound source identification. By extracting sound features and computing correlations between different sound sources, the system creates an intermediate representation that enables distinction between desired and interference signals even when they arrive from the same direction, thus resolving the contradiction between noise reduction and sound source distinction.
2Reliability
If beamforming forms a radiation pattern for a specific source, then signal retention for that source is improved, but other sound sources within the same radiation pattern are not filtered out
Solution Approach 1:
The patent applies local quality by extracting and analyzing specific sound features (such as spectral characteristics, temporal patterns, or frequency content) that are unique to each sound source. Instead of treating all signals within the radiation pattern uniformly, the system identifies and preserves only those signals that match the local characteristics of the desired source while filtering out signals with different feature patterns, even if they are spatially close.
Solution Approach 2:
The patent replaces the purely spatial filtering mechanism of traditional beamforming with a feature-based correlation mechanism. Instead of relying solely on the physical radiation pattern to separate sources, the system substitutes this with computational correlation analysis of sound features, enabling distinction between sources based on their acoustic characteristics rather than just spatial position.
3Productivity
If multiple sound sources are present in front of the microphone array, then recording completeness is improved, but signal processing complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-extracting sound features and pre-computing correlation matrices for multiple sound sources before the actual signal separation is needed. The system prepares feature representations and correlation data in advance, which then can be quickly applied during real-time processing to separate multiple sound sources without requiring complex real-time computation, thus reducing processing complexity while maintaining recording completeness.
Data Source
AI summary
A sound receiving processing method and a sound signal processing apparatus are provided. In the method, multiple input sound signals are received through a microphone array. The input sound signals are combined into a main sound signal based on beamforming. Multiple primary extraction signals belonging to the main sound signal are respectively extracted from the input sound signals. Multiple estimated sound source signals corresponding to the sound sources are separated from the primary extraction signals through blind signal separation. Multiple re-extraction signals belonging to the estimated sound source signals are respectively extracted from the primary extraction signals. A distance of the first source relative to the microphone array is determined according to a power ratio. One of the estimated sound source signals corresponding to the first source is determined according to a distance limit. Therefore, the sound signal from the desired sound source could be obtained.