Hearing Device Noise Reduction via Frequency-Dependent Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Hearing aids face challenges in providing a natural hearing experience with effective noise reduction, as existing methods like beamforming and noise filtering can introduce artifacts and fail to accurately distinguish sound sources, especially in situations with rapid spectral changes.
Innovation Solution
A method involving a hearing device that receives inside and outside channel sound signals from microphones, determines a desired signal and a disturb signal, calculates a frequency band-dependent weight factor by dividing the magnitude of the desired signal by the sum of the desired and disturb signal magnitudes, and adjusts the inside channel sound signal with this weight factor to reduce noise, while maintaining spectral acoustic information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If beamforming is used to amplify sound from specific directions, then the signal-to-noise ratio is improved, but the device complexity increases and spectral acoustic information is lost
Solution Approach 1:
The patent divides the noise reduction task into two independent parts: (1) beamforming applied only to outside channel microphones for noise estimation, and (2) simple spectral subtraction applied to the inside channel microphone signal. This segmentation reduces overall complexity compared to applying beamforming to all microphones while maintaining noise reduction effectiveness.
Solution Approach 2:
The patent moves the complex beamforming operation from the time domain to the frequency domain, where noise estimation can be performed more efficiently. By transforming the outside channel signals into frequency components, the system can estimate noise spectrum without requiring complex time-domain beamforming operations on all microphone signals.
2Reliability
If noise filtering is applied to remove disturb signals, then the signal-to-noise ratio is improved, but artifacts like musical noise are generated that reduce sound quality
Solution Approach 1:
The patent applies noise reduction only partially - using spectral subtraction on the inside channel signal rather than aggressive filtering. By applying the noise estimate selectively and with controlled subtraction, the system achieves noise reduction while minimizing the generation of musical noise artifacts that would result from more aggressive filtering approaches.
Solution Approach 2:
The patent uses the outside channel microphones as intermediaries to estimate the noise spectrum. Instead of directly filtering the inside channel signal, the system first captures ambient noise through outside microphones, processes this intermediate signal to estimate the noise spectrum, and then uses this estimate to reduce noise in the primary inside channel signal.
3Reliability
If behind-the-ear hearing aids are used with multiple microphones for beamforming, then noise reduction is improved, but spectral acoustic information from the concha is lost
Solution Approach 1:
The patent extracts the noise estimation function from the main signal path by using separate outside channel microphones. The inside channel microphone, which preserves spectral acoustic information from the concha, is used only for capturing the desired sound signal with its natural spectral characteristics. Noise estimation is performed separately using outside microphones, and the two processed signals are combined, thus extracting noise reduction functionality without compromising the spectral information in the primary signal path.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
A method for reducing noise in a sound signal of a hearing device (10) comprises: receiving an inside channel sound signal (Si) from a first microphone (12) in an ear channel of a user, a first outside channel sound signal (SO1) from a second microphone (14) outside of the ear channel and a second outside channel sound signal (SO2) from a third microphone (16) outside of the ear channel; determining a desired signal (SDes) from the first and second outside channel sound signals (SO1, SO2) and a disturb signal (SDis) from the first and second outside channel sound signals; determining a frequency band dependent weight factor (ω) by dividing a magnitude value of the desired signal (SDes) in a frequency band through at least a magnitude value of the disturb signal (SDis) in the frequency band; and generating a noise reduced sound signal (SA) by multiplying the inside channel sound signal (SI) with the weight factor (ω).