Auxiliary-Conditioned Neural Networks for Adaptive Hearing Aid Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hearing aids use fixed neural networks that do not adapt to individual hearing aid users or varying acoustic situations, necessitating impractical training processes that are infeasible for on-the-go adjustments.
Innovation Solution
A hearing aid system with adaptively adjustable neural network weights, enabled by configuration data from an auxiliary device, allowing for personalized and situational signal processing without extensive training at the clinic.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If fixed neural networks are used in hearing aids, then device complexity is reduced and ease of manufacture is improved, but adaptability to individual users and acoustic situations deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-training neural networks for different acoustic situations and user profiles before deployment. Multiple pre-trained neural networks are stored in memory, and the system selects and loads the appropriate pre-trained network based on detected acoustic conditions and user characteristics, avoiding the need for real-time training while achieving adaptability.
Solution Approach 2:
The system implements dynamics by making the neural network configuration changeable based on operating conditions. The hearing aid dynamically selects different pre-trained neural networks according to acoustic environment and user needs, and can update neural network parameters in real-time based on feedback, transforming a static system into an adaptive one.
2Adaptability or versatility
If individualized neural network training is performed for each user, then adaptability is improved, but loss of time and productivity deteriorate due to lengthy training processes
Solution Approach 1:
The patent resolves this contradiction by performing the time-consuming neural network training in advance during manufacturing or initial setup. Multiple pre-trained neural networks covering different user profiles and acoustic situations are prepared beforehand and stored in memory. During actual use, the system only needs to select and load the appropriate pre-trained network, reducing adaptation time from hours/days to seconds.
Solution Approach 2:
The system creates copies of trained neural networks for different user profiles and acoustic situations. Instead of training a new network for each user interaction, pre-trained network copies are stored and selectively activated. This allows rapid deployment of individualized processing without repeating the training process.
3Adaptability or versatility
If multiple pre-trained neural networks are stored and selected based on acoustic situations, then adaptability is improved, but device complexity and memory requirements increase
Solution Approach 1:
The patent applies segmentation by dividing the acoustic environment into distinct situations (e.g., quiet, noisy, reverberant) and creating separate pre-trained neural networks for each situation. The system segments the overall adaptation problem into manageable categories, storing specialized networks for each segment rather than attempting to create one universal network.
Solution Approach 2:
The hearing aid system achieves universality by designing a framework that can handle multiple acoustic situations and user profiles through a common architecture. The same hardware platform and software framework support various pre-trained neural networks, allowing the system to be universally applicable across different conditions without requiring separate dedicated systems for each situation.
Data Source
AI summary
A hearing aid adapted to be worn in or at an ear of a hearing aid user and/or to be fully or partially implanted in the head of the hearing aid user is disclosed. The hearing aid comprises a processing unit connected to said input unit and to said output unit, where the processing unit comprises a neural network, and where the processing unit is configured to determine signal processing parameters of the hearing aid based on weights of the neural network. A hearing system and a corresponding method is furthermore disclosed.