Hearing Device Neural Network Processing for Noisy Speech
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Hearing devices struggle to enhance target sounds, such as speech, in noisy environments, as simple amplification does not effectively distinguish target sounds from environmental noise, leading to difficulty in speech intelligibility.
Innovation Solution
A hearing system employs a neural network-based nonlinear signal processing algorithm to process microphone signals, controlling microphone directionality and enhancing target sounds like speech in noisy environments by improving the signal-to-noise ratio without substantial distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If simple amplification is used to enhance target sounds, then the volume of target sounds is increased, but the ability to distinguish target sounds from environmental noise is not improved
Solution Approach 1:
The patent segments the audio signal processing into multiple independent stages: initial amplification stage, neural network processing stage for noise separation, and final output stage. This segmentation allows the system to amplify signals while subsequently separating target speech from noise, resolving the contradiction between amplification gain and speech intelligibility.
Solution Approach 2:
The neural network acts as an intermediary component between the amplification stage and the final output. It processes the amplified signal to separate target speech from environmental noise, enabling the system to maintain both high amplification gain and clear speech intelligibility by mediating the signal transformation.
2Device complexity
If traditional linear signal processing is used, then the system is simple to implement, but it cannot achieve sufficient noise reduction in noisy environments
Solution Approach 1:
The patent replaces traditional linear mechanical signal processing with a neural network-based nonlinear processing system. This substitution enables significantly better noise reduction and speech enhancement capabilities, achieving superior signal-to-noise ratio improvement despite the increased computational complexity.
Solution Approach 2:
The system changes the fundamental processing parameter from linear to nonlinear operations through the neural network. This parameter change enables the system to adaptively process signals in noisy environments, achieving much better noise reduction performance compared to traditional linear methods.
3Loss of information
If neural network-based nonlinear processing is used to improve signal-to-noise ratio, then speech intelligibility is enhanced, but computational complexity increases
Solution Approach 1:
The neural network is pre-trained offline with large datasets to learn optimal speech enhancement patterns. This preliminary action transfers complex computational knowledge into the network's weight parameters, allowing the device to achieve high signal-to-noise ratio improvement with relatively simple real-time inference operations during actual use.
Data Source
AI summary
A hearing system performs nonlinear processing of signals received from a plurality of microphones using a neural network to enhance a target signal in a noisy environment. In various embodiments, the neural network can be trained to improve a signal-to-noise ratio without causing substantial distortion of the target signal. An example of the target sound includes speech, and the neural network is used to improve speech intelligibility.