Hearing Sound Processing with ML-Tuned Parameters and Low Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional hearing devices employ a one-size-fits-all approach for sound processing programs, leading to suboptimal parameterization that does not account for individual user anatomy, behavior, or environmental context, resulting in degraded performance.
Innovation Solution
Implementing a machine learning model that dynamically optimizes sound processing program parameters in real-time by leveraging a trained model to adjust coefficients for IIR filters based on user and environmental information, using a dual-path processing approach with low-latency and long-latency paths to enhance stability and adaptivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If fixed parameters are used for sound processing programs, then device complexity is reduced and ease of operation is improved, but adaptability to individual user anatomy, behavior, and environmental context deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-training a machine learning model offline using extensive user data, anatomy information, and environmental context. This pre-training phase performs the complex computational work beforehand, allowing the hearing device to then use the trained model for real-time parameter optimization without requiring complex onboard computation during actual operation.
Solution Approach 2:
The patent introduces a machine learning model as an intermediary between raw sensor data and sound processing parameters. This intermediary component processes the complex relationships between user anatomy, behavior patterns, and environmental factors, translating them into optimized parameters for beamformers, noise canceling, and target gain programs without requiring direct complex processing in the hearing device itself.
2Adaptability or versatility
If real-time dynamic optimization of sound processing parameters is implemented, then adaptability to environmental conditions and user behavior is improved, but processing latency increases
Solution Approach 1:
The machine learning model is pre-trained offline with extensive data before deployment in the hearing device. This preliminary training phase performs the computationally intensive work of learning complex patterns, allowing the model to make rapid predictions in real-time with minimal processing latency during actual use.
Solution Approach 2:
The system dynamically adjusts sound processing parameters in real-time based on changing environmental conditions and user behavior patterns. The machine learning model continuously processes current sensor data to optimize parameters adaptively, allowing the system to respond to dynamic conditions while maintaining low latency through efficient inference.
3Reliability
If fixed sound processing programs are used, then reliability and stability of processing are improved, but performance optimization for specific situations deteriorates
Solution Approach 1:
The patent implements dynamic parameter adjustment within a stable processing framework. The machine learning model continuously adapts parameters based on real-time conditions while maintaining a reliable underlying processing architecture. This allows the system to optimize performance for specific situations without compromising the overall stability and reliability of sound processing.
Solution Approach 2:
The system changes processing parameters dynamically based on environmental conditions and user behavior patterns. By adjusting parameters such as beamformer weights, noise canceling coefficients, and target gain settings according to real-time inputs, the system optimizes performance for different situations while maintaining stable processing through the structured application of these parameter changes.
4Productivity
If machine learning model processing is added to optimize parameters, then adaptability and performance are improved, but device complexity and computational requirements increase
Solution Approach 1:
The complex computational work of training the machine learning model is performed preliminarily and offline, before deployment in the hearing device. This allows the model to be trained on extensive datasets without requiring complex real-time computation capabilities in the hearing device itself, reducing the computational burden during actual operation.
Solution Approach 2:
The machine learning model serves as an intermediary processing layer that simplifies the overall system architecture. Instead of implementing complex rule-based systems or multiple separate processing modules, the trained model consolidates the optimization logic into a single component that takes sensor inputs and produces optimized parameters, reducing overall system complexity despite the advanced processing it enables.
Data Source
AI summary
An exemplary hearing device includes a memory that stores instructions and a processor communicatively coupled to the memory and configured to execute the instructions to perform a process. The process may comprise processing an audio signal in accordance with a sound processing program along a first signal processing path, the sound processing program configured to compensate for individual hearing loss of a user of the hearing device; identifying information associated with the audio signal; providing the information associated with the audio signal to a trained machine learning model that processes the information along a second signal processing path, the trained machine learning model configured to output one or more parameters that are optimized on the fly for the sound processing program based on the information; and applying the one or more parameters output from the trained machine learning model to the sound processing program.