Hearing Device Signal Processing Using Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hearing aids and cochlear implants fail to provide effective signal processing that accurately compensates for individual hearing impairments due to limited electrodes and non-specific stimulation, leading to degraded sound perception and difficulty in hearing speech in noise.
Innovation Solution
A method using supervised machine learning, specifically deep neural networks, to define and set nonlinear signal processing for hearing devices by training on normal and impaired auditory representations, adjusting parameters to minimize deviation between input and output signals, and incorporating supra-threshold measures for individualized compensation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional linear signal processing is used in hearing aids, then the device complexity is low, but the sound perception quality and speech intelligibility in noise are degraded
Solution Approach 1:
The patent transforms the linear signal processing parameters into non-linear parameters that adapt to individual hearing impairments. The system uses non-linear compression and expansion functions that dynamically adjust gain based on input level and frequency, thereby improving sound perception quality while managing device complexity through parameter optimization.
Solution Approach 2:
The patent implements dynamic signal processing where the hearing aid continuously adapts its transfer function based on real-time acoustic environment and user-specific hearing loss characteristics. This dynamic adaptation allows the system to optimize sound perception quality across varying conditions without requiring overly complex fixed processing for all scenarios.
2Device complexity
If cochlear implants use a limited number of electrodes, then the device complexity is reduced, but the spectral resolution and sound perception are severely degraded
Solution Approach 1:
The patent applies segmentation by dividing the auditory spectrum into multiple frequency channels that are processed independently through non-linear transformations. Each electrode channel is segmented to handle specific frequency ranges with customized non-linear compression, allowing the limited electrode array to achieve better spectral resolution by optimally distributing and processing frequency information across channels.
3Reliability
If auditory models are highly nonlinear to capture physiological hearing processes, then the sound perception accuracy is improved, but the ability to invert the model for signal processing is lost
Solution Approach 1:
The patent introduces an intermediary non-linear transformation stage between the linear signal processing and the auditory model. This intermediary layer uses learned non-linear functions that bridge the gap, allowing the system to maintain accurate auditory representation while enabling inversion through the intermediary transformation. The intermediary acts as a mediator that preserves model accuracy while facilitating signal processing adaptability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of defining and setting a nonlinear signal processing of a hearing device, e.g. a hearing aid, by machine learning is provided. The hearing device being configured to be worn by a user at or in an ear or to be fully or partially implanted in the head at an ear of the user, the method comprising providing at least one electric input signal representing at least one input sound signal from an environment of a hearing device user, determining a normal-hearing representation of said at least one electric input signal based on a normal-hearing auditory model, determining a hearing-impaired representation of said at least one electric input signal based on a hearing-impaired auditory model, determining optimised training parameters by machine learning, where determining optimised training parameters comprises iteratively adjusting the training parameters, and comparing the normal-hearing representation with the hearing-impaired representation to determine a degree of matching between the normal-hearing representation and the hearing-impaired representation, until the degree of matching fulfils predetermined requirements, and, when the degree of matching fulfils the predetermined requirements, determining corresponding signal processing parameters of the hearing device based on the optimised training parameters. A hearing device is further provided.