Sound Source Detection Using Multi-Class Pattern Recognition

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing sound source detection systems face challenges in accurately detecting the direction and distance of approaching vehicles at high speeds, due to limited spatial resolution and reliance on threshold values, which can lead to variable detection accuracy and difficulty in determining whether the vehicle is approaching or receding.

Innovation Solution

A sound source detection system using a multi-class pattern recognition method to classify sound characteristics and estimate the position of the sound source, eliminating the need for threshold values and improving detection accuracy by setting classes based on time intervals, distances, and moving directions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Length of stationary object

If the distance between microphones is increased to improve maximum detection distance, then the detection range is extended, but spatial resolution deteriorates and aliasing occurs

Engineering Contradiction:
Improvedistance between microphonesVSAvoidspatial resolution
Core Design Contradiction:
Length of stationary objectVSMeasurement precision

Solution Approach 1:

The patent divides the detection space into multiple zones based on delay of arrival values, creating distinct detection regions (first detection region, second detection region, third detection region) with different threshold values. This segmentation allows the system to maintain high detection accuracy across varying distances without requiring a fixed microphone spacing, thereby resolving the contradiction between extended detection range and spatial resolution.

Inventive Principle:
Principle #1Segmentation

2Device complexity

If a fixed threshold value is used for vehicle detection, then the detection process is simplified, but detection accuracy varies significantly

Engineering Contradiction:
Improvedetection process complexityVSAvoiddetection accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent implements dynamic threshold adjustment based on the calculated delay of arrival value. The threshold value changes according to the detection region (first, second, or third region), allowing the system to adapt to different sound source distances and conditions. This dynamic approach maintains high detection accuracy while adding minimal complexity to the detection process.

Inventive Principle:
Principle #15Dynamics

3Loss of information

If delay of arrival method is used to determine approaching direction, then direction information can be obtained, but it cannot determine whether the vehicle is approaching or receding from a single detection

Engineering Contradiction:
Improvedirection informationVSAvoidapproach determination reliability
Core Design Contradiction:
Loss of informationVSReliability

Solution Approach 1:

The patent uses the delay of arrival value as feedback to determine both the detection region and the approach state. By comparing the delay of arrival against predefined thresholds, the system can reliably determine whether a vehicle is approaching or receding, in addition to identifying its direction. This feedback mechanism resolves the contradiction by extracting multiple pieces of information from the same measurement.

Inventive Principle:
Principle #23Feedback

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 system achieves high accuracy in detecting approaching vehicles by classifying sound characteristics and estimating their position, providing reliable detection regardless of the vehicle's speed and direction, with improved stability even in noisy conditions.

Implementation Method 1

a sound collecting device that collects sound

Methodology Applied
Scientific EffectMicrophone transduction:

Data Source

PatentEP2758793B1Sound source detection system
Publication Date: 2016.10.26 TOYOTA JIDOSHA KK
  • EP2758793B1 patent drawingFigure 1
  • EP2758793B1 patent drawingFigure 2
  • EP2758793B1 patent drawingFigure 3

AI summary

A sound source detection system for detecting a sound source (e.g., running sound of a vehicle) based on sound collected by microphones extracts characteristic amounts from the sound collected by the microphones, sets a plurality of classes according to the position of the sound source by a multi-class pattern recognition method (e.g., multi-class SVM) using the characteristic amounts, extracts characteristic amounts from sound collected by the microphones for detection of a sound source, determines a class to which the extracted characteristic amounts belong, from the preset plurality of classes, and estimates the sound source based on the class.