Eigenvector Sound Source Localization for Stationary Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sound source localization methods, such as the MUSIC method, fail to accurately estimate sound source orientation in systems with stationary noise sources, particularly in moving robots with rotating heads, as they assume stronger sound sources than noise, leading to incorrect localization and the need for multiple correlation matrices based on head rotation.
Innovation Solution
A sound source localization apparatus and method using eigenvectors, which calculates eigenvalues from correlation matrices, including a correction unit to adjust predetermined correlation matrices based on microphone posture information, reducing the impact of stationary noise and allowing correct localization without pre-storing various correlation matrices for different head positions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the MUSIC method is used for sound source localization, then the method is easier to detect peaks of the spatial spectrum, but the sound source orientation cannot be estimated correctly when stationary noise sources have large power
Solution Approach 1:
The patent changes the fundamental parameter used for localization from eigenvalue-based (MUSIC method) to eigenvector-based approach. By using eigenvectors of the correlation matrix combined with predetermined correlation matrices representing noise models, the system can distinguish between sound sources and noise even when noise power is high, resolving the contradiction between measurement precision and noise influence.
2Adaptability or versatility
If multiple correlation matrices are prepared for different head rotation positions, then sound source localization can be performed for various postures, but a lot of data must be stored
Solution Approach 1:
The patent creates a universal solution by using a single correlation matrix calculation unit that processes correlation matrices for all head postures through a unified eigenvector-based algorithm. Instead of storing multiple separate correlation matrices, the system uses one correlation matrix and combines it with predetermined noise correlation matrices, reducing data storage while maintaining versatility across different head positions.
3Reliability
If the MUSIC method assumes sound sources are stronger than noise, then the method works for typical conditions, but the assumption is not satisfied in systems with stationary noise sources having large power
Solution Approach 1:
The patent introduces predetermined correlation matrices as intermediaries that represent noise models. These predetermined matrices act as mediators between the actual measured correlation matrix and the noise components, allowing the system to separate sound sources from noise even when noise power exceeds sound source power, thus improving reliability in high noise environments.
Data Source
AI summary
A sound source localization apparatus for localizing a sound source using an eigenvector, includes, a sound signal input unit inputting a sound signal, a correlation matrix calculation unit calculating a correlation matrix of the input sound signal, and an eigenvector calculation unit calculating an eigenvalue of the correlation matrix using the calculated correlation matrix, wherein the eigenvector calculation unit calculates the eigenvector using the correlation matrix of the input sound signal and one or more predetermined correlation matrices.