Hearing Aid Adaptive Filter Control for Feedback and Tonal Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hearing devices face challenges in effectively controlling the adaptation speed of adaptive filters to efficiently reduce feedback noise, leading to either slow adaptation or misadapations, especially with tonal input signals, which affect sound quality and distinguishability between feedback whistling and external tonal sounds.
Innovation Solution
A method for controlling the adaptation step size of an adaptive filter in hearing devices by recording sound signals, obtaining autocorrelation values, and adjusting the frequency shift to decorrelate signals, allowing for optimal adaptation speed in sub-bands while minimizing artefacts, and reducing the step size only when autocorrelation values are near unity, indicating tonal signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If the adaptation step size is increased to speed up adaptation to feedback path changes, then the adaptation speed is improved, but misadaptations occur with tonal input signals degrading sound quality
Solution Approach 1:
The adaptation step size is made dynamic by adjusting it according to the detected signal type. The system automatically increases the step size when feedback whistling is detected and decreases it when tonal input signals are present, allowing optimal performance in different operating conditions without manual intervention.
Solution Approach 2:
The system changes the adaptation parameter (step size) based on the autocorrelation value of the microphone signal. By monitoring the autocorrelation and comparing it to a threshold, the system adapts the step size parameter to match the current signal characteristics, resolving the contradiction between speed and reliability.
2Reliability
If the adaptation step size is decreased to prevent misadaptations with tonal signals, then sound quality is maintained, but the adaptation speed becomes too slow to respond to feedback path changes
Solution Approach 1:
The system dynamically adjusts the adaptation step size based on real-time detection of signal characteristics. When tonal input signals are detected through autocorrelation analysis, the step size is reduced to maintain sound quality. When feedback whistling is detected, the step size is increased to ensure rapid adaptation, thus resolving the speed-reliability trade-off.
Solution Approach 2:
The adaptation parameter (step size) is changed based on the autocorrelation value of the microphone signal. The system monitors the autocorrelation and adjusts the step size accordingly - decreasing it for tonal signals to prevent misadaptations and increasing it for feedback whistling to maintain fast adaptation, thereby resolving the contradiction.
3Measurement precision
If autocorrelation analysis is performed with zero time difference to detect tonal signals, then detection accuracy is improved, but feedback whistling cannot be distinguished from external tonal sounds
Solution Approach 1:
The autocorrelation analysis is segmented into multiple time difference values rather than using a single zero time difference. By analyzing autocorrelation at different time lags, the system can distinguish between feedback whistling (which exhibits specific autocorrelation patterns at non-zero lags) and external tonal sounds, resolving the detection ambiguity.
Solution Approach 2:
The time difference parameter in autocorrelation analysis is changed from a fixed zero value to multiple variable values. The system evaluates autocorrelation at different time lags and uses this information to distinguish feedback whistling from external tonal signals, improving both detection accuracy and distinguishability simultaneously.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
Feedback whistling and external tonal signals should be better distinguished during feedback suppression. Therefore, a method is proposed for controlling the adaptation step size of an adaptive filter (8) of a hearing device for feedback reduction by picking up a sound signal (n+r) through a microphone (2) of the hearing device and outputting a corresponding microphone signal (m), generating a listener signal (x') for a listener (4) of the hearing device based on the microphone signal (m), and reducing a feedback signal (r) in the microphone signal (m) by the adaptive filter (8). For this purpose, an autocorrelation value (a) is obtained from samples of the microphone signal (m) between which there is a time difference (k), and the adaptation step size (p) of the adaptive filter (8) is controlled based on the autocorrelation value (a).A frequency of an output signal (x) obtained on the basis of the microphone signal (m) is shifted when generating the listener signal (x') and the time difference (k) for obtaining the autocorrelation value (a) is controlled as a function of the shift of the frequency (9) of the microphone signal (m).