Hearing Aid Speaker Tracking via Speech Profile Database
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current hearing aids face challenges in distinguishing between desired speech and background noise, particularly in scenarios where multiple sound sources are present, leading to reduced effectiveness in noisy environments and requiring the user to constantly focus on the desired speaker.
Innovation Solution
A method that utilizes a database of speech profiles to selectively amplify and prioritize the acoustic signal of known speakers through blind source separation, allowing the hearing aid to track and emphasize preferred speaker sources independently of their spatial location.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If blind source separation algorithms are used to separate multiple sound sources, then the ability to separate speech from background noise is improved, but the ability to distinguish between desired speech and background noise in the same frequency region deteriorates
Solution Approach 1:
The patent introduces a direction identification unit as an intermediary component that determines the spatial direction of reference persons using speaker recognition. This directional information serves as a mediator between the blind source separation process and the final signal selection, enabling the system to resolve frequency region ambiguities by using spatial cues to distinguish desired speech from background noise in overlapping frequency bands.
Solution Approach 2:
The patent adds a spatial dimension to the frequency-based signal processing by incorporating directional information from the direction identification unit. While blind source separation operates in the frequency domain, the system extends analysis to the spatial domain by determining reference person directions, thereby creating a multi-dimensional approach that resolves ambiguities in frequency region overlap.
2Measurement precision
If directional microphone algorithms are used to focus on specific directions, then speech intelligibility from preferred directions is improved, but the system becomes ambiguous when multiple competing sources are present at the same location
Solution Approach 1:
The direction identification unit acts as an intermediary that resolves ambiguities in multi-source scenarios by determining the spatial direction of reference persons through speaker recognition. This directional mediation enables the system to distinguish between competing speakers at similar locations by analyzing their spatial characteristics, thereby maintaining speech intelligibility without excessive complexity.
Solution Approach 2:
The system incorporates feedback through the direction identification unit that continuously monitors and updates the spatial information of reference persons. This feedback mechanism enables dynamic adjustment of the directional focus based on the current acoustic scene, allowing the system to adapt to changing conditions and resolve ambiguities in real-time.
3Measurement precision
If noise reduction algorithms are applied to reduce background noise, then the signal-to-noise ratio is improved, but the algorithms cannot distinguish between speech and background noise in the same frequency region
Solution Approach 1:
The direction identification unit serves as an intermediary that provides spatial information to distinguish between speech and background noise in overlapping frequency regions. By determining the direction of reference persons through speaker recognition, the system can selectively enhance signals from desired directions while suppressing noise from other directions, even when frequency spectra overlap.
Solution Approach 2:
The patent extends noise reduction from the frequency domain to the spatial domain by incorporating directional information. This dimensional expansion allows the system to differentiate between speech and noise based on their spatial characteristics rather than solely relying on frequency spectral analysis, thereby improving reliability in frequency region distinction.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method for operating a hearing aid (1), wherein, for tracking and selecting a speaker acoustic source (102; s1(t), sn(t)) from an ambient sound (100; 102, 104; s1(t), s2(t), ..., sn(t)), a "speaker" operating mode is established by a signal processing (300) of the hearing aid (1), wherein the hearing aid (1) generates electrical acoustic signals (202, 212; 312, 314; 322, 324; 332; x1(t), x2(t), ..., xn(t); s'1(t), s'2(t), ...) from the recorded ambient sound (100; 102, 104; s(t), s2(t), ..., sn(t))., s'n(t)) are generated, from which an electrical speaker signal (322; s'1(t), s'n(t)) is identified and selected by the signal processing (300) through a database (340) with speech profiles (P) of preferred speakers, wherein the electrical speaker signal (322; s'1(t), s'n(t)) is selectively taken into account in an output sound (402; s"(t); s"1(t)+s"n(t)) of the hearing aid (1) such that it is at least acoustically prominent for the hearing aid wearer in comparison with another acoustic source (104; s2(t)) and is thus perceived better by the hearing aid wearer.