Neural Network Hearing Aid Processing for Noisy Speech
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional hearing aids struggle to effectively separate speech from background noise in noisy environments due to limitations in computational power and the impracticality of incorporating neural networks, leading to difficulties in maintaining an optimal signal-to-noise ratio and user experience.
Innovation Solution
A dual-path signal processing chain in hearing aids that selectively engages neural network-based audio enhancement (NNE) and digital signal processing (DSP), with a neural network engine (NNE) integrated into the signal path to enhance audio before playback, and a controller determining the processing path based on user inputs, environmental factors, and sensor data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If neural network-based audio enhancement is incorporated into the hearing aid signal path, then speech separation from background noise is improved, but device complexity and computational power requirements increase
Solution Approach 1:
The patent divides the audio processing into two distinct paths: a neural network processing path for speech enhancement and a traditional DSP path for other audio processing. This segmentation allows the complex neural network to be isolated to only when needed, reducing overall system complexity while maintaining speech separation accuracy.
Solution Approach 2:
The patent implements dynamic path selection where the controller automatically switches between neural network processing and traditional DSP based on environmental conditions, signal characteristics, and power availability. This dynamic approach optimizes the balance between speech separation performance and computational complexity in real-time.
2Measurement precision
If neural network processing is continuously applied, then speech intelligibility in noisy environments is improved, but power consumption increases
Solution Approach 1:
Instead of continuous neural network processing, the patent implements periodic activation based on environmental assessment. The system periodically evaluates noise levels and speech presence, activating neural network processing only during periods when speech enhancement is actually needed, thereby conserving battery power while maintaining speech intelligibility when required.
Solution Approach 2:
The patent incorporates feedback mechanisms where the controller continuously monitors environmental conditions and audio signal characteristics, then adjusts neural network activation accordingly. This feedback loop ensures power-efficient operation by activating the computationally intensive neural network only when the acoustic environment warrants speech enhancement.
3Device complexity
If traditional signal processing techniques are used, then device simplicity is maintained, but signal-to-noise ratio improvement is insufficient in noisy environments
Solution Approach 1:
The patent introduces an intermediary controller that manages two processing paths: a simple traditional DSP path for normal conditions and a sophisticated neural network path for noisy environments. This intermediary component allows the system to maintain simplicity when possible while accessing advanced speech separation capabilities when needed, effectively bridging the gap between processing simplicity and SNR improvement.
Solution Approach 2:
The patent creates a composite processing architecture that combines traditional digital signal processing methods with modern neural network techniques. This hybrid approach leverages the strengths of both methodologies: the efficiency and simplicity of traditional DSP for general audio processing, and the superior speech separation capabilities of neural networks for noisy environments.
Data Source
AI summary
The disclosure generally relates to a method, system and apparatus to improve a user's understanding of speech in real-time conversations by processing the audio through a neural network contained in a hearing device. The hearing device may be a headphone or hearing aid. In one embodiment, the disclosure relates to an apparatus to enhance incoming audio signal. The apparatus includes a controller to receive an incoming signal and provide a controller output signal; a neural network engine (NNE) circuitry in communication with the controller, the NNE circuitry activatable by the controller, the NNE circuitry configured to generate an NNE output signal from the controller output signal; and a digital signal processing (DSP) circuitry to receive one or more of controller output signal or the NNE circuitry output signal to thereby generate a processed signal; wherein the controller determines a processing path of the controller output signal through one of the DSP or the NNE circuitries as a function of one or more of predefined parameters, incoming signal characteristics and NNE circuitry feedback.