Sound Source Detection Using Correlation Matrix Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sound source detection methods face challenges in accurately detecting a target sound source when noise levels are higher than the sound source, making it difficult to calculate a correlation matrix corresponding only to noise components.
Innovation Solution
A sound source detection apparatus that calculates a first correlation matrix from acoustic signals, specifies a non-scan range to exclude noise sources, estimates a second correlation matrix based on direction vectors and a spatial spectrum, and removes this matrix from the first correlation matrix to obtain a third correlation matrix corresponding to the target sound source within a scan range, allowing for accurate localization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a correlation matrix is calculated from observed acoustic signals to detect sound sources, then sound source direction can be estimated, but noise components with higher sound pressure levels than the target sound source contaminate the correlation matrix and prevent accurate detection
Solution Approach 1:
The patent segments the correlation matrix calculation into two distinct parts: (1) calculating a first correlation matrix from observed signals containing both noise and target sound source, and (2) calculating a second correlation matrix from signals in non-scan ranges where only noise is present. This segmentation allows separate processing of noise and target components, enabling the target sound source to be isolated by subtracting the noise-only correlation matrix from the total correlation matrix.
Solution Approach 2:
The patent extracts the noise component from the total signal by calculating a correlation matrix specifically from signals in non-scan ranges (directions where the target sound source is not present). This extracted noise correlation matrix is then removed from the total correlation matrix to obtain a clean correlation matrix containing only the target sound source information, effectively taking out the harmful noise component.
2Adaptability or versatility
If the scan range is expanded to cover all possible directions, then complete sound source detection is attempted, but detection accuracy decreases when noise sources are present in multiple directions
Solution Approach 1:
The patent introduces dynamic range management by defining both scan ranges (where target sound source detection is attempted) and non-scan ranges (where only noise is present). The system dynamically adjusts which directions are processed for target detection versus which directions are used purely for noise estimation, allowing flexible adaptation to different operational scenarios while maintaining high detection accuracy.
Solution Approach 2:
The patent uses non-scan ranges as an intermediary mechanism to estimate noise characteristics without being contaminated by the target sound source. These non-scan ranges serve as a mediator that provides clean noise data, which is then used to process the scan range data and improve target detection accuracy.
Data Source
AI summary
A first correlation matrix, which corresponds to observed signals, is calculated, a non-scan range is specified, a second correlation matrix, which corresponds to an acoustic signal from a sound source within the non-scan range, is estimated, a third correlation matrix, which corresponds to a target sound source within a scan range, is calculated by removing the second correlation matrix from the first correlation matrix, and a first spatial spectrum, which is a localization result, is calculated from the third correlation matrix. In the estimation, the second correlation matrix is estimated from direction vectors calculated from a direction range of the non-scan range and a second spatial spectrum, which is a localization result calculated immediately before the first spatial spectrum is calculated.