Sound Source Localization via Fourier Series Steering Vector Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional sound source localization methods, such as beam forming and MUSIC, require significant computational resources to calculate steering vectors for each discrete angle, leading to increased calculation time and inefficiency.
Innovation Solution
The method employs Fourier series expansion to model steering vectors for sound signals in the frequency domain, allowing for the calculation of steering vectors for arbitrary angles using Fourier base functions and coefficients, reducing the computational burden by selecting a model order N that minimizes the product (M+K)(2N+1) compared to (M×K).
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If steering vectors are calculated for each discrete angle using conventional beam forming or MUSIC methods, then sound source localization accuracy is maintained, but the computational complexity and calculation time increase significantly
Solution Approach 1:
The patent transforms the steering vector calculation from spatial domain to frequency domain by applying Fourier series expansion. Instead of calculating steering vectors for each discrete angle in the spatial domain, the method models steering vectors using frequency domain representations with Fourier coefficients, fundamentally changing the parameter space and reducing computational complexity while maintaining localization accuracy
Solution Approach 2:
The patent replaces the conventional mechanical calculation approach (direct computation of steering vectors for each angle) with a mathematical transformation approach (Fourier series expansion). This substitution eliminates the need for repeated calculations at discrete angles by using continuous frequency domain modeling, significantly reducing the computational burden
2Productivity
If steering vectors are pre-calculated for each discrete angle, then sound source localization can be performed, but the preparation time and computational resources are excessive
Solution Approach 1:
The patent performs preliminary action by pre-calculating only the Fourier coefficients in the frequency domain, which can then be used to generate steering vectors for any arbitrary angle without re-calculating from scratch. This preliminary computation of frequency domain parameters enables rapid generation of steering vectors when needed, reducing both preparation time and real-time calculation time
Solution Approach 2:
The Fourier series expansion models steering vectors in a universal manner that works for any arbitrary angle, not just discrete angles. The same frequency domain model can generate steering vectors for any angle in the continuous domain, making the calculation method universally applicable and eliminating the need for separate calculations at each discrete angle point
Data Source
AI summary
A sound source localization device includes: a sound receiving unit that includes two or more microphones; and a sound source localization unit that transforms a sound signal received by each of the microphones into a frequency domain, models a steering vector through Fourier series expansion of an N-th (here, N is an integer equal to or larger than “1”) order for the transformed sound signal of the frequency domain for each of the microphones, calculates a steering vector of an arbitrary angle using the modeled steering vector, and performs localization of a sound source using the calculated steering vector of the arbitrary angle.