Hearing Aid Speaker Identification via Facial Recognition

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In real-time audio signal processing for hearing aids, particularly in complex acoustic environments, it is challenging to automatically and reliably amplify speech contributions from preferred conversation partners relative to other signal contributions, leading to decreased speech intelligibility and user inconvenience in situations like 'cocktail party' scenarios.

Innovation Solution

A method involving an auxiliary device that generates image recordings to recognize preferred conversation partners through facial recognition, analyzing audio sequences for characteristic speaker identification parameters, and adjusting signal processing to highlight their contributions relative to others, using a database for evaluation and processing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If complex non-real-time signal processing algorithms are used to amplify speech contributions, then speech intelligibility is improved, but real-time processing capability deteriorates

Engineering Contradiction:
Improvespeech intelligibilityVSAvoidreal-time processing capability
Core Design Contradiction:
ReliabilityVSSpeed

Solution Approach 1:

The patent segments the audio signal processing into distinct functional modules: direction-of-arrival estimation module, speaker identification module, and speech enhancement module. Each module processes specific aspects of the signal independently, enabling real-time operation while maintaining high speech intelligibility through specialized processing for each segment.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If manual control of signal processing modes is required for individualized amplification, then amplification precision is improved, but ease of operation deteriorates

Engineering Contradiction:
Improveamplification precisionVSAvoidease of operation
Core Design Contradiction:
Measurement precisionVSEase of operation

Solution Approach 1:

The system performs self-service by automatically identifying preferred conversation partners through facial recognition and voice pattern matching, then automatically applying individualized amplification settings. The hearing aid monitors the acoustic environment continuously and adjusts signal processing parameters without requiring user intervention, thereby maintaining high amplification precision while significantly improving ease of operation.

Inventive Principle:
Principle #25Self-service

3Adaptability or versatility

If frequent activation and changing of signal processing modes is required, then adaptability to different situations is improved, but loss of time increases

Engineering Contradiction:
Improveadaptability to different situationsVSAvoidtime for mode switching
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The system performs preliminary action by pre-identifying and storing voice patterns of preferred conversation partners in advance. When a preferred partner is detected in the acoustic environment, the system immediately retrieves the stored profile and applies the appropriate signal processing configuration without requiring mode switching or user intervention, thereby maintaining high adaptability while eliminating time loss associated with mode changes.

Inventive Principle:
Principle #10Preliminary action

4Extent of automation

If automatic speaker identification is implemented, then extent of automation is improved, but device complexity increases

Engineering Contradiction:
Improveautomatic speaker identificationVSAvoiddevice complexity
Core Design Contradiction:
Extent of automationVSDevice complexity

Solution Approach 1:

The patent implements multi-functionality by integrating multiple capabilities into a unified system: the auxiliary device performs both facial recognition and voice pattern recording, while the hearing aid combines direction-of-arrival estimation, speaker identification, and speech enhancement in a single integrated signal processing chain. This universal approach achieves high automation while managing device complexity through shared hardware and software resources across multiple functions.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP3693960B1Method for individualized signal processing of an audio signal of a hearing aid
Publication Date: 2024.09.25 SIVANTOS PTE LTD
  • EP3693960B1 patent drawingFigure 1
  • EP3693960B1 patent drawingFigure 2

AI summary

The invention describes a method for individualized signal processing of an audio signal (12) from a hearing aid, wherein in a recognition phase (1) a first image recording (8) is generated by an auxiliary device (4), the presence of a preferred conversation partner (10) is inferred from the first image recording (8), and a first audio sequence (14) of the audio signal (12) and/or an auxiliary audio signal of the auxiliary device (4) is then analyzed for characteristic speaker identification parameters (30), and the speaker identification parameters (30) determined in the first audio sequence (14) are stored in a database (31), and wherein in an application phase (40) the audio signal (12) is analyzed with respect to the stored speaker identification parameters (30), and thereby evaluated with regard to the presence of the preferred conversation partner (10).and, upon detection of the presence of the preferred conversation partner (10), their signal contributions in the audio signal (12) are amplified.