Sound Source Tracking via Dual Microphone Array Integration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sound source localization techniques lack robustness and accuracy in tracking sound sources, particularly when multiple sound sources are present and their speeds vary, leading to ambiguity in localization.
Innovation Solution
A sound source tracking system that integrates measurements from a moving-body microphone array and a fixed microphone array using a particle filter, with a sub-array method to reduce computational load and a transition model that adapts based on sound source speed, and weighted likelihood integration for improved accuracy and robustness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple microphone arrays are used for sound source localization, then measurement accuracy and robustness are improved, but computational complexity and processing time increase
Solution Approach 1:
The system divides the sound source localization task into two independent measurement streams: one using a moving-body microphone array for directional information and another using a fixed microphone array for positional information. Each processor handles one array independently, then results are integrated. This segmentation reduces computational complexity while maintaining high accuracy by leveraging the strengths of each array type.
Solution Approach 2:
The system merges the directional measurement results from the moving-body microphone array with the positional measurement results from the fixed microphone array through integration processing. This combining approach consolidates the complementary information from both arrays to achieve superior localization accuracy and robustness that neither array could provide alone.
2Measurement precision
If all microphones in the fixed microphone array are used for calculation, then measurement completeness is improved, but computational load increases
Solution Approach 1:
The system applies local quality by selectively using only those microphones from the fixed array that are spatially close to the sound source, rather than uniformly processing signals from all microphones. This localized approach concentrates computational resources on the most relevant sensors, maintaining measurement accuracy while significantly reducing overall computational load.
3Adaptability or versatility
If a single transition model is used in the particle filter, then system simplicity is maintained, but adaptability to varying sound source speeds deteriorates
Solution Approach 1:
The system implements dynamics by making the transition model in the particle filter adaptive to sound source speed variations. Rather than using a static, single transition model, the system dynamically adjusts the transition model based on detected sound source speed, enabling accurate tracking of both stationary and moving sound sources while managing complexity through conditional model selection.
Data Source
AI summary
A result of a sound source direction measurement based on an output of an REMA (first microphone array) (11) and a result of a sound source position measurement based on an output of an IRMA (second microphone array) (12) are integrated through a particle filter or in space. Thus, the different microphones, i.e., the REMA (11) and the IRMA (12) can cancel mutual defects or ambiguities with each other. Therefore, from views of improvement in accuracy and robustness a performance of sound source localization can be improved.