Machine Learning Earpiece Acoustic Mass Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the optimal acoustic mass of earpieces struggle with complexity and accuracy, often leading to user rejection due to issues like feedback, occlusion effect, and performance degradation, particularly when considering asymmetric hearing losses.
Innovation Solution
A machine learning algorithm, specifically an artificial neuronal network, is trained using user audiograms and additional data to predict the optimal acoustic mass of earpieces, simplifying the selection process and improving accuracy by considering user-specific factors such as hearing asymmetry and experience.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If rule-based algorithms are used to determine optimal acoustic mass, then a balance between rejection metrics can be achieved, but the algorithm complexity becomes high and accuracy is limited
Solution Approach 1:
The patent replaces complex rule-based algorithms with a machine learning model that has been trained on user data and acoustic mass parameters. This substitution transforms the deterministic, complexity-heavy rule-based system into a data-driven model that achieves higher accuracy without requiring complex real-time calculations, thereby resolving the contradiction between measurement precision and device complexity.
2Adaptability or versatility
If rule-based algorithms consider only a subset of user characteristics, then algorithm complexity is reduced, but user acceptance decreases due to asymmetric hearing loss not being addressed
Solution Approach 1:
The patent changes the input parameters of the determination system by incorporating multiple user characteristics including monaural hearing loss, asymmetry metrics, and experience levels into the machine learning model. This comprehensive parameter set allows the system to adapt to individual user needs, particularly addressing asymmetric hearing loss, thereby improving user acceptance and reliability without increasing algorithmic complexity.
3Manufacturing precision
If traditional methods determine acoustic mass, then manufacturing processes remain simple, but precision of acoustic mass determination is insufficient leading to feedback and occlusion effects
Solution Approach 1:
The patent applies preliminary action by training the machine learning model in advance with comprehensive user data and acoustic mass relationships. This pre-trained model can then quickly determine optimal acoustic mass values during the fitting process, achieving high manufacturing precision without adding complexity to the actual manufacturing process. The complex work is done beforehand during model training, not during production.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
A method for determining an acoustic mass (26) of an earpiece (14) to be plugged into an ear of a user comprises: receiving user data (22) comprising at least an audiogram of the user; and inputting the user data (22) into a machine learning algorithm (24) and determining the acoustic mass (26) by the machine learning algorithm (24).