Hearing Aid Gain Margin Prediction for Feedback Onset Control

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing hearing aid technologies struggle to predict and prevent feedback distortions efficiently, often requiring excessive computational resources and being less effective in detecting the onset of feedback, leading to audible distortions such as howling or chirping.

Innovation Solution

A deep neural network is trained to predict the instantaneous gain margin in hearing aids using signals from microphones, receivers, and filter coefficients, allowing for adaptive adjustment of the step-size of the adaptive feedback canceller and gain to prevent feedback distortions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional feedback detection methods are used in hearing aids, then feedback distortions can be detected, but the computational resources required are excessive and the detection effectiveness is reduced

Engineering Contradiction:
Improvefeedback detection effectivenessVSAvoidcomputational resources
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent changes the parameter being monitored from complex feedback signal analysis to a simpler gain margin parameter. By training a neural network to predict gain margin directly from basic input signals (microphone and receiver signals), the system achieves reliable feedback detection while reducing computational complexity. The neural network learns to predict this critical stability parameter during training, enabling efficient real-time operation in the hearing device.

Inventive Principle:
Principle #35Parameter changes

2Reliability

If gain margin prediction is implemented to prevent feedback, then feedback distortions are reduced, but additional computational overhead is introduced

Engineering Contradiction:
Improvefeedback prevention capabilityVSAvoidcomputational overhead
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system performs preliminary action by training the neural network offline to predict gain margin before deploying it in the hearing device. During training, the network learns from synthesized data the relationship between input signals and gain margin. Once trained, the compact model requires minimal computational resources during real-time operation, enabling proactive feedback prevention without excessive energy consumption.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent replaces traditional mechanical or algorithmic feedback detection methods with a neural network-based prediction system. This substitution allows the system to predict gain margin and potential feedback conditions before they occur, enabling preventive action rather than reactive correction, thereby reducing the computational overhead associated with continuous complex signal analysis.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Object-generated harmful factors

If adaptive feedback cancellation is used, then feedback distortions can be reduced, but the system becomes less effective at detecting the onset of feedback

Engineering Contradiction:
Improvefeedback distortionsVSAvoidfeedback onset detection
Core Design Contradiction:
Object-generated harmful factorsVSDifficulty of detecting and measuring

Solution Approach 1:

The system uses feedback in the form of neural network predictions about future gain margin conditions. By continuously predicting gain margin based on current and past signals, the system creates a feedback mechanism that warns of impending feedback conditions before they manifest as audible distortions. This predictive feedback enables early detection and preventive action.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The neural network performs preliminary analysis of signal patterns that precede feedback onset. During training, the network learns to recognize precursors to feedback conditions by predicting gain margin trends. This preliminary detection capability allows the system to take preventive action before feedback distortions occur, improving both feedback reduction and onset detection effectiveness.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4287659B1Predicting gain margin in a hearing device using a neural network
Publication Date: 2025.12.24 STARKEY LABORATORIES INC
  • EP4287659B1 patent drawingFigure 1
  • EP4287659B1 patent drawingFigure 2
  • EP4287659B1 patent drawingFigure 3

AI summary

A hearing device includes a microphone that produces an audio input signal and a loudspeaker that outputs an amplified audio signal into an ear canal. A signal processing path is coupled to the microphone and the loudspeaker. The signal processing path includes a deep neural network configured to predict an instantaneous gain margin of the hearing device based on a set of inputs. The set of inputs includes a first parameter of the audio input signal, a second parameter of the amplified audio signal, and a gain of the signal processing path. A feedback reduction module of the device receives the predicted instantaneous gain margin and adjusts feedback reduction parameters to reduce an onset of feedback in the hearing device