Hearing Aid Sound Classification Using Hierarchical Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing hearing aid systems face challenges in providing precise and robust sound classification using minimal processing resources, which can lead to deteriorated sound quality and speech intelligibility due to complex and resource-intensive classification methods.

Innovation Solution

A hearing aid system that employs a sound environment classifier with a feature extractor providing Mel Frequency Cepstral Coefficients, modulation cepstrum coefficients, and tonality features, using a band-pass filter bank with non-linearly spaced frequency bands and a classifier comprising a base class and final class classifier to efficiently classify sound environments with reduced processing requirements.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If complex sound classification methods are used to improve sound environment identification accuracy, then measurement precision is improved, but device complexity increases and processing resources are consumed

Engineering Contradiction:
Improvesound environment classification accuracyVSAvoidclassification system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The sound environment classification system is segmented into multiple hierarchical levels: a first level with multiple base classes (including music class) and a second level with more specific sound environment classes. This segmentation allows the system to achieve high classification accuracy by progressively narrowing down from broad categories to specific environments, while keeping each individual classification stage relatively simple and computationally efficient.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a hierarchical dimension to the classification system, organizing sound environments in multiple levels from general to specific. This dimensional approach transforms a potentially complex single-level classification problem into a series of simpler multi-level decisions, improving accuracy without proportionally increasing overall system complexity.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Measurement precision

If complex sound classification methods are used to improve sound environment identification accuracy, then measurement precision is improved, but processing resources are consumed

Engineering Contradiction:
Improvesound environment classification accuracyVSAvoidprocessing resource consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

By segmenting the classification into hierarchical levels where the music class is identified at an intermediate stage, the system avoids the computational burden of processing all possible sound environment categories simultaneously. This segmentation enables efficient resource usage by focusing processing power on distinguishing music from other environments before proceeding to more specific classifications.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system performs preliminary classification into base classes including music before proceeding to final sound environment classification. This preliminary action of identifying music early in the hierarchy allows subsequent processing to be optimized or simplified, reducing overall processing resource consumption while maintaining high accuracy in music environment detection.

Inventive Principle:
Principle #10Preliminary action

3Device complexity

If simple classification methods are used to reduce processing resources, then device complexity is reduced, but measurement precision deteriorates

Engineering Contradiction:
Improveclassification system complexityVSAvoidsound environment classification accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The hierarchical segmentation allows each classification level to remain relatively simple in structure while collectively achieving high precision. By dividing the classification task into manageable stages with clear decision boundaries, the system maintains low complexity at each step while the cumulative effect of multiple stages delivers high overall classification accuracy for sound environments including music.

Inventive Principle:
Principle #1Segmentation

4Use of energy by moving object

If simple classification methods are used to reduce processing resources, then processing resources are conserved, but measurement precision deteriorates

Engineering Contradiction:
Improveprocessing resource consumptionVSAvoidsound environment classification accuracy
Core Design Contradiction:
Use of energy by moving objectVSMeasurement precision

Solution Approach 1:

The preliminary classification into base classes serves as an efficient filtering stage that consumes minimal processing resources while providing valuable information for subsequent classification. This preliminary action of grouping sound environments into broader categories first allows the system to achieve high final classification accuracy without the computational expense of analyzing all detailed categories simultaneously, thus conserving processing resources.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3182729B1Hearing aid system and a method of operating a hearing aid system
Publication Date: 2019.11.06 WIDEX AS
  • EP3182729B1 patent drawingFigure 1~2
  • EP3182729B1 patent drawingFigure 3
  • EP3182729B1 patent drawingFigure 4

AI summary

A method of operating a hearing aid system (400) based on a classification of the current sound environment, which includes a measure of a beat probability and a hearing aid system for carrying out such a method.