Hearing Aid Speech Extraction via Central Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Hearing impaired individuals face significant challenges in understanding speech in noisy environments due to reverberation and multiple dynamic interferences, with existing monaural or binaural microphone arrays and noise reduction algorithms providing limited benefits in real-world scenarios.
Innovation Solution
A system comprising hearing aids with microphones that extract the wearer's own voice using near-field array processing, wirelessly transmitting these signals to a central processing station for enhancement and mixing based on individual hearing loss and preferences, and then sending the mixed signal back to the hearing aid for playback, utilizing a full-duplex wireless link to reduce computational and power constraints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If monaural or binaural microphone arrays are used to improve speech understanding in noisy environments, then speech understanding is improved, but the benefits are limited due to reverberation and multiple dynamic interferences
Solution Approach 1:
The patent segments the audio signal processing into multiple independent stages: initial noise reduction, voice activity detection, spectral estimation, and iterative refinement. Each stage processes specific aspects of the signal separately, allowing targeted optimization of speech enhancement while minimizing the impact of reverberation and interference through sequential rather than simultaneous processing
Solution Approach 2:
The patent performs preliminary noise reduction and spectral estimation before main speech enhancement processing. By pre-processing the signal to remove obvious noise components and estimate the noise spectrum in advance, the system creates a cleaner input for subsequent speech enhancement stages, thereby improving overall speech understanding in reverberant environments
2Measurement precision
If noise reduction algorithms are used to improve speech understanding, then speech understanding is improved, but computational complexity and power consumption increase
Solution Approach 1:
The patent applies noise reduction algorithms selectively based on voice activity detection rather than continuously processing all audio signals. By detecting when speech is present and applying enhancement only during those periods, the system achieves effective speech understanding improvement while significantly reducing computational complexity and power consumption during non-speech intervals
Solution Approach 2:
The system uses periodic voice activity detection to determine when to activate intensive noise reduction processing. This periodic activation pattern allows the computationally expensive algorithms to run only when necessary (during speech segments), thereby maintaining speech understanding quality while managing power consumption and computational load efficiently
3Measurement precision
If complex signal processing is applied to extract speech signals in noisy environments, then speech intelligibility is improved, but device complexity increases
Solution Approach 1:
The complex signal processing is divided into modular functional blocks: noise reduction module, voice activity detection module, spectral estimation module, and speech enhancement module. Each module handles a specific aspect of the processing chain, making the overall complex system more manageable, easier to implement, and allowing independent optimization of each component while maintaining high speech intelligibility
Solution Approach 2:
The patent introduces intermediate processing stages including noise spectrum estimation and voice activity detection results as intermediary data structures. These intermediaries bridge the input noisy signal and the final enhanced output, allowing complex transformations to be broken down into manageable steps that can be implemented with standard signal processing components
Data Source
AI summary
Methods and devices are described for improves speech understanding by hearing aid users in crowded environments. In one embodiment, a hearing aid wearer's speech signal is extracted using the microphone or microphones in the hearing aid. The hearing aid is configured to wirelessly transmit the extracted speech signals to a central processing station that enhances the extracted speech signals received from all registered hearing aids. The central processing station processes the received speech signals individually based on provided hearing losses mixes the signals based on the provided preferences, wirelessly transmits the mixed signal to each registered hearing aid for play back.