Hearing Aid Neural Network Tuning Across Sound Levels

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing hearing aids with fixed neural networks perform suboptimally across varying sound input levels and auditory model channels, leading to erratic behavior in noise reduction and hearing loss compensation.

Innovation Solution

A method is introduced to optimize training parameters for neural networks in hearing aids, using a selected normal-hearing and hearing-impaired auditory model to equalize sound pressure levels, thereby improving performance across sound input levels and auditory model channels without increasing computational load.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If a fixed neural network is used for all hearing aid users and acoustic situations, then the device complexity is reduced and ease of manufacture is improved, but the performance and reliability deteriorate across varying sound input levels and auditory model channels

Engineering Contradiction:
Improveneural network configurationVSAvoidperformance consistency
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

The patent implements dynamic adaptation of neural network parameters based on acoustic situation and user characteristics. The system transitions from a fixed neural network to one that dynamically adjusts its parameters (such as gain, compression ratios, and frequency weighting) according to real-time acoustic conditions and individual user needs, thereby maintaining reliability across varying conditions without substantially increasing device complexity

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes the parameters of the neural network based on acoustic situation and user data. Specifically, it adjusts parameters such as noise reduction thresholds, hearing loss compensation factors, and signal processing gains according to the acoustic environment and individual user characteristics, allowing the same neural network architecture to perform optimally across diverse conditions

Inventive Principle:
Principle #35Parameter changes

2Adaptability or versatility

If training data spans a large dynamic input range for hearing loss compensation, then the adaptability to different acoustic situations is improved, but the manufacturing precision and performance across sound levels deteriorate due to erratic neural network behavior

Engineering Contradiction:
Improveacoustic situation coverageVSAvoidperformance consistency
Core Design Contradiction:
Adaptability or versatilityVSManufacturing precision

Solution Approach 1:

The patent applies local quality by optimizing neural network parameters for specific acoustic situations and sound input levels rather than attempting a single universal configuration. It divides the acoustic space into different regions (quiet, moderate, noisy environments) and applies locally optimized parameters to each region, ensuring high performance consistency within each acoustic condition while maintaining broad adaptability across all conditions

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent performs preliminary training and optimization of neural network parameters across a comprehensive dataset of acoustic situations before deployment. By pre-adjusting parameters for various acoustic conditions and user types during the manufacturing process, the system ensures consistent performance across diverse operating conditions without requiring complex real-time adaptation mechanisms

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4513902A1Improved hearing loss emulation via neural networks
Publication Date: 2025.02.26 OTICON
  • EP4513902A1 patent drawingFigure 1
  • EP4513902A1 patent drawingFigure 2
  • EP4513902A1 patent drawingFigure 3

AI summary

A method of defining and setting a signal processing of a hearing aid is disclosed. The hearing aid is configured to be worn by a user at or in an ear of the user. The method comprises providing at least one electric input signal representing at least one input sound signal from a sound environment of a hearing aid user, determining a normal-hearing representation of said at least one electric input signal based on a selected normal-hearing auditory model fj, determining optimised training parameters of a neural network, where the neural network represents a hearing-impaired representation of said at least one electric input signal based on a hearing-impaired auditory model, wherein determining the optimised training parameters comprises determining a frequency distribution, βj, and a level and frequency distribution, αj,1, of said at least one electric input signal based on an equalization of sound pressure levels of said at least one electric input signal. A hearing aid adapted to be worn in or at an ear of a user is furthermore disclosed.