Bone Conduction Speech Reconstruction for Noisy Hearing Aids
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing bone conduction sensors in hearing devices suffer from deficiencies such as muffled sound, missing higher frequencies, and sensor noise, leading to poor signal quality and signal-to-noise ratio in noisy environments.
Innovation Solution
Employ a synthetic speech model trained using bone conduction signals and ambient microphone signals to generate high-quality synthetic speech, allowing the model to reconstruct speech without requiring ambient microphone input during operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If a bone conduction sensor is used to capture speech signals, then environmental noise interference is reduced, but the speech quality becomes muffled and higher frequencies are lost
Solution Approach 1:
The patent combines bone conduction sensor signals with ambient microphone signals through hybrid signal processing. The bone conduction sensor captures speech with minimal environmental noise, while the ambient microphone preserves higher frequencies and natural speech characteristics. By merging these complementary signals, the system achieves both noise rejection and high-quality speech reproduction.
Solution Approach 2:
The patent creates a composite speech signal by integrating two different signal sources: bone conduction vibrations and air-conducted sound waves. This composite approach leverages the strengths of each source - the noise immunity of bone conduction and the frequency fidelity of air conduction - to produce a speech signal that overcomes the limitations of either source alone.
2Measurement precision
If a directional microphone or boom-microphone is used to reduce environmental noise, then signal-to-noise ratio improves to a certain extent, but device complexity increases
Solution Approach 1:
The patent uses bone conduction sensors to capture a copy of the user's speech signal through vibrations transmitted through the skull. This alternative copying method bypasses the need for complex directional microphone arrangements, as the bone conduction path naturally isolates the user's voice from environmental noise without requiring sophisticated spatial filtering or multiple microphone elements.
3Object-affected harmful factors
If bone conduction signal is used directly for speech transmission, then environmental noise has little influence, but the signal includes artefacts from swallowing, jaw-movements, and sensor noise
Solution Approach 1:
The patent extracts the useful speech signal from the bone conduction sensor output by separating it from unwanted artefacts. Through signal processing techniques, the system identifies and extracts the speech-related vibrations while filtering out artefacts from swallowing, jaw movements, and sensor noise. The extracted clean speech signal is then combined with ambient microphone input for final output.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The synthetic speech model effectively enhances speech quality by approximating air-conducted voice sounds, improving signal-to-noise ratio and reducing noise interference, particularly in noisy environments.
Implementation Method 1
a bone conduction sensor configured to convert bone vibrations of voice sound information into a bone conduction signal
Data Source
Figure 1A
Figure 1B
Figure 2A
AI summary
The present disclosure relates to a hearing apparatus comprising: a bone conduction sensor configured to convert bone vibrations of voice sound information into a bone conduction signal; a signal processing unit configured to implement a synthetic speech generation process, the synthetic speech generation process implementing a speech model; wherein the synthetic speech generation process receives the bone conduction signal as a control input and outputs a synthetic speech signal.