Microphone Array DOA Estimation via Spectral Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating the directions of arrival (DOAs) of multiple simultaneous acoustic sources using microphone arrays are inaccurate and computationally intensive, making them unsuitable for real-time applications in speech interfaces and noise-cancelling devices.
Innovation Solution
A computer-implemented method that bandpass-filters audio signals into spectral bands, computes cross-products between microphone pairs, and iteratively minimizes errors between cross-product sets and model cross-product sets to estimate DOAs. This method distinguishes between multiple acoustic sources and measures their individual DOAs with increased precision, while maintaining low computational requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional cross-correlation methods are used to estimate DOA for multiple simultaneous sources, then measurement precision may be maintained, but computational complexity increases significantly making real-time processing infeasible
Solution Approach 1:
The patent segments the DOA estimation problem by dividing the set of active sources into multiple subsets, where each subset is processed independently through dedicated processing paths. This allows parallel computation of DOA estimates for different source groups, reducing the overall computational burden while maintaining accurate measurement through multiple independent estimation channels.
Solution Approach 2:
The patent computes DOA estimates for multiple subsets of active sources, where each subset represents a partial computation. By computing estimates for several subsets simultaneously and then selecting the most reliable estimate (e.g., based on confidence metrics or consistency checks), the system achieves accurate DOA measurement without performing the full computationally intensive calculation for all possible source combinations.
2Measurement precision
If refined approaches are used to distinguish multiple simultaneous sources, then measurement precision improves, but device complexity and processing requirements increase
Solution Approach 1:
The patent divides the complex task of separating multiple simultaneous sources into multiple simpler tasks by creating separate processing paths for different subsets of sources. Each processing path independently analyzes signals from a specific subset, making the separation problem more tractable and computationally efficient while maintaining high precision through the collective results of multiple specialized processors.
3Measurement precision
If exhaustive computation is performed to achieve accurate DOA estimates, then measurement precision is maximized, but processing time increases making real-time applications impractical
Solution Approach 1:
The patent computes DOA estimates for multiple subsets of active sources rather than performing exhaustive computation for all possible source configurations. By calculating estimates for several subsets in parallel and selecting the best result based on confidence metrics, the system achieves sufficiently accurate DOA estimation in real-time without the prohibitive computational cost of exhaustive analysis.
Solution Approach 2:
The patent segments the computation into multiple independent subset-processing paths that can be executed simultaneously. This parallelization of computational tasks reduces the overall processing time required to achieve accurate DOA estimates, making real-time processing feasible while maintaining measurement precision through the coordinated results of multiple segmented computations.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The method achieves precise DOA estimation for multiple simultaneous acoustic sources with reduced computational needs, enabling real-time applications in various audio systems without the need for expensive processing hardware.
Implementation Method 1
Cross-correlation between the signals from microphone pairs is the conventional approach to estimate said time differences
Data Source
AI summary
A computer-implemented method for measuring the directions of arrival, DOAs, of acoustic signals recorded by a microphone array outputting audio signals, comprising: bandpass-filtering the audio signals into at least one spectral band; in each spectral band, for at least one pair of microphones, computing a cross-product between the audio signals of said pair, iteratively minimizing an error between the cross-products and model cross-products calculated from DOA estimates, and determining those DOA estimates whose powers constitute local maxima, as DOA candidates; and measuring the DOAs of the acoustic signals from the DOA candidates of all spectral bands. Further disclosed is an apparatus for performing this method.