Hearing Aid Acoustic Transfer Function Database
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current hearing aid systems face challenges in accurately estimating absolute or relative acoustic transfer functions (AATFs/RATFs) for proper beamforming and noise reduction, particularly in adapting to individual user head characteristics and microphone positions, leading to suboptimal sound processing.
Innovation Solution
A hearing aid system with a processor and database that uses a dictionary of frequency-dependent ATF-vectors to determine personalized ATF-vectors based on user-specific acoustic propagation models, incorporating multiple microphones and accounting for various head orientations and positions to enhance sound localization and noise reduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a dictionary method is used to estimate acoustic transfer functions, then the estimation can be performed, but the computational complexity when searching the dictionary is high
Solution Approach 1:
The patent segments the acoustic transfer function estimation by dividing the dictionary search into physically plausible subsets based on head characteristics and microphone positions. Instead of searching the entire dictionary, the system divides it into manageable segments that correspond to specific user anatomies and device placements, reducing computational complexity while maintaining estimation accuracy.
Solution Approach 2:
The patent applies local quality by tailoring the dictionary search to individual users' specific characteristics. Each user receives a personalized subset of the dictionary that matches their head anatomy and microphone positions, rather than using a generic full dictionary. This localized approach improves estimation precision for each user while reducing the overall computational burden.
2Adaptability or versatility
If a generic acoustic transfer function model is used, then the system is simpler to implement, but it cannot adapt to individual user head characteristics and microphone positions
Solution Approach 1:
The patent applies preliminary action by pre-computing and storing acoustic transfer functions for various head characteristics and microphone positions in a dictionary before actual use. During operation, the system simply needs to select the appropriate pre-computed functions from the dictionary that match the user's characteristics, avoiding complex real-time calculations while achieving personalized adaptation.
Solution Approach 2:
The patent utilizes parameter changes by varying the acoustic transfer function parameters according to user-specific characteristics such as head size, shape, and microphone positions. The system selects and applies different parameters from the dictionary based on measured or estimated user characteristics, enabling adaptation to individual users without requiring a completely different system for each user.
3Measurement precision
If the full dictionary is searched for ATF-vectors, then the most accurate match can be found, but the search time and computational resources increase
Solution Approach 1:
The patent extracts only the relevant portion of the dictionary that is physically plausible for a given user based on their head characteristics and microphone positions. By taking out and removing irrelevant entries from the full dictionary, the system maintains the ability to find the most accurate match while significantly reducing the search space and associated time requirements.
Data Source
AI summary
A hearing aid microphone system includes M microphones providing corresponding electric input signals. Environmental sound at a given microphone includes a target sound signal propagated via an acoustic propagation channel from a direction to or a location of a target sound source to the microphone, and possible additive noise signals. The acoustic propagation channel is modeled. The hearing aid system includes: a processor connected to the microphones, and a database Θ having a multitude of dictionaries Δp, p=1, . . . , P, where p is a person index, of vectors, termed ATF-vectors, whose elements ATFm, m=1, . . . , M, are frequency dependent acoustic transfer functions representing direction- or location-dependent, and frequency dependent propagation of sound. The processor is configured to, at least in a learning mode of operation, determine personalized ATF-vectors based on the multitude of dictionaries Δp, the electric input signals, and the model of the acoustic propagation channels.