Hearing Aid Speech Intelligibility via ML Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current hearing aids primarily focus on amplifying sound frequencies and reducing ambient noise but fail to effectively enhance speech intelligibility, particularly in noisy environments, and do not adequately address the challenge of discerning specific language or spoken words for individuals with hearing impairments.
Innovation Solution
A customizable hearing aid system that uses machine learning to process speech signals, learning from individual user data and broader demographic populations, to enhance speech intelligibility by transforming input speech articulations into more intelligible outputs, taking into account sensorineural aspects and brain auditory processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If prior art hearing aids amplify sound frequencies and reduce ambient noise, then sound intensity is improved, but speech intelligibility remains insufficient
Solution Approach 1:
The patent replaces traditional mechanical/acoustic signal processing methods with machine learning-based artificial intelligence systems. The hearing aid uses trained neural networks to analyze and transform speech signals, converting them into articulations that are more intelligible for the specific user's hearing profile, thereby resolving the contradiction between amplifying sound intensity and maintaining speech intelligibility.
Solution Approach 2:
The system dynamically changes multiple parameters of the speech signal simultaneously, including frequency distribution, amplitude modulation, and temporal characteristics, based on the user's audiogram data and hearing impairment profile. This multi-parameter transformation approach enables the system to enhance speech intelligibility while accounting for the user's specific hearing deficiencies.
2Device complexity
If hearing aids process speech signals using traditional methods, then device complexity is low, but adaptability to individual hearing profiles is insufficient
Solution Approach 1:
The system performs preliminary actions by collecting and processing user-specific hearing data during fitting sessions, creating personalized audiograms and training custom machine learning models for each user. This preliminary customization enables the hearing aid to adapt to individual hearing profiles with high precision, resolving the contradiction between processing complexity and adaptability.
Solution Approach 2:
The hearing aid system continuously learns and adapts to the user's hearing characteristics through ongoing data collection and model refinement. The system serves itself by automatically adjusting its processing parameters based on real-world usage patterns and feedback, reducing the need for manual reconfiguration while maintaining high adaptability to the user's evolving hearing needs.
3Ease of manufacture
If hearing aids amplify all frequencies equally, then ease of manufacture is high, but speech discernment in noisy environments is poor
Solution Approach 1:
The patent applies local quality by customizing the signal processing for each user's specific hearing deficiencies. Instead of uniform amplification across all frequencies, the system identifies and targets specific frequency ranges and speech components that are most problematic for each individual user, applying enhanced processing only where needed. This resolves the contradiction by maintaining manufacturing simplicity while dramatically improving speech discernment through personalized frequency-specific processing.
Data Source
AI summary
A hearing aid system presents a hearing impaired user with customized enhanced intelligibility sound in a preferred language. The system includes a model trained with a set of source speech data representing sampling from a speech population relevant to the user. The model is also trained with a set of corresponding alternative articulation of source data, pre-defined or algorithmically constructed during an interactive session with the user. The model creates a set of selected target speech training data from the set of alternative articulation data that is preferred by the user as being satisfactorily intelligible and clear. The system includes a machine learning model, trained to shift incoming source speech data to a preferred variant of the target data that the hearing aid system presents to the user.