Hearing Aid Dynamic Beamforming for Multi-Talker Speech Intelligibility
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current hearing assistance technologies fail to enhance speech intelligibility and reduce listening effort in multi-talker settings, as they rely on single-microphone noise reduction algorithms and directional systems that require a constant target talker direction, which is not feasible in dynamic scenarios like dinner conversations.
Innovation Solution
A hearing assistance system that senses and estimates the facing orientation, location, and talking activity of talkers using sensors and a processor, adjusting device parameters to enhance speech intelligibility by identifying and tracking the most promising target listening direction, even in changing environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If directional systems are used to increase speech intelligibility by passing signals from the direction of a target talker, then speech intelligibility is improved, but the system fails when the target talker direction changes dynamically
Solution Approach 1:
The patent implements dynamic target tracking by continuously monitoring the spatial positions of multiple talkers and automatically adjusting the directional beamforming to follow the most relevant talker. The system transitions from static directional assumptions to real-time adaptive directionality, allowing the beam to dynamically rotate and reposition as speakers move or take turns talking.
Solution Approach 2:
The system employs active feedback mechanisms where the hearing aid continuously receives spatial audio information, identifies which talker is currently being addressed by the user (through gaze direction or response patterns), and adjusts the directional enhancement accordingly. This closed-loop feedback enables the system to adapt to changing conversation dynamics without requiring manual user input.
2Ease of operation
If single-microphone noise reduction algorithms are used to reduce listening effort, then listening effort is reduced, but speech intelligibility is not improved in multiple-talker settings
Solution Approach 1:
The patent combines the advantages of both single-microphone noise reduction and multi-microphone directional beamforming into a unified processing pipeline. The system first applies noise reduction algorithms to suppress background noise and competing speech, then layers directional enhancement on top to selectively amplify the target talker. This merged approach achieves both reduced listening effort and improved speech intelligibility simultaneously.
Solution Approach 2:
The audio processing is segmented into distinct functional stages: noise reduction processing separates desired speech from background noise, while directional beamforming separately enhances the spatially-located target talker. This segmentation allows each processing stage to optimize for its specific function without interfering with the other, achieving both noise reduction and speech intelligibility improvement.
3Measurement precision
If static directionality systems are used to pass signals from the direction of a target talker, then speech intelligibility is increased, but the system requires the target talker to remain in a constant position
Solution Approach 1:
The system replaces static directional assumptions with dynamic target tracking. Multiple microphones continuously capture spatial audio information, and the system calculates the positions of all active talkers in real-time. The directional beamforming parameters are dynamically updated to follow the most relevant talker, allowing the enhancement to remain effective even as speakers move around the conversation space.
Data Source
AI summary
Disclosed herein, among other things, are systems and methods for hearing assistance in multiple-talker settings. One aspect of the present subject matter includes a method of operating a hearing assistance device for a user in an environment. A parameter is sensed relating to facing orientation of a talker in communication within the environment. Parameters related to location and talking activity of a talker can also be used. In various embodiments, facing orientation, location, and talking activity of the talker are estimated based on the sensed parameter. A hearing assistance device parameter is adjusted based on the estimated facing orientation, location, and talking activity of the talker, according to various embodiments.