Hearing Device Speech Intelligibility Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current hearing aid technologies face challenges in assessing speech intelligibility in noisy multi-talker environments, as existing methods require a 'clean' reference speech signal, which is often unavailable in real-life situations, leading to inaccurate intelligibility assessments.
Innovation Solution
A hearing device with a processor and controller that decomposes input signals into frequency domain representations using characterization blocks, allowing for non-intrusive estimation of speech intelligibility without a reference signal, enabling the reconstruction of a reference speech signal for improved assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If intrusive methods (STOI, NCM) are used to assess speech intelligibility, then measurement precision is improved, but device complexity increases and reference signal availability is required
Solution Approach 1:
The patent extracts only the essential spectral characteristics needed for intelligibility assessment from the complex signal processing chains of STOI and NCM. By taking out only the critical frequency domain representations and simplifying the processing steps, the system achieves comparable measurement precision with reduced computational complexity and without requiring reference signals.
Solution Approach 2:
The system uses the degraded speech signal itself to assess intelligibility without requiring external reference signals. The method processes the single available input signal through spectral analysis and intelligibility metrics, making the system self-sufficient and suitable for real-world hearing aid applications where reference signals are unavailable.
2Measurement precision
If reference speech signal is required for intelligibility assessment, then measurement precision is improved, but ease of operation deteriorates due to signal unavailability
Solution Approach 1:
The system performs self-assessment of speech intelligibility using only the degraded speech signal as input. By eliminating the requirement for reference signals, the system becomes operationally simple and automatically adapts to real listening conditions, while maintaining accurate intelligibility measurement through spectral analysis and optimized processing algorithms.
3Ease of operation
If non-intrusive estimation without reference signal is implemented, then ease of operation is improved, but measurement precision deteriorates
Solution Approach 1:
The patent transforms the degraded speech signal into the frequency domain and applies specific parameter optimizations to enhance intelligibility metric accuracy. By changing the representation parameters (spectral resolution, frequency weighting) and applying correction factors, the system achieves high measurement precision despite working with only the degraded signal without reference.
Solution Approach 2:
The system performs preliminary spectral analysis and signal characterization before computing the intelligibility metric. By preprocessing the signal to extract essential spectral features and apply appropriate weighting functions in advance, the system ensures accurate measurement precision is maintained even without reference signals.
Data Source
AI summary
A hearing device includes: an input module for provision of a first input signal; a processor configured to provide an electrical output signal based on the first input signal; a receiver configured to provide an audio output signal; and a controller comprising a speech intelligibility estimator configured to determine a speech intelligibility indicator indicative of speech intelligibility based on the first input signal, wherein the controller is configured to control the processor based on the speech intelligibility indicator; wherein the speech intelligibility estimator comprises a decomposition module configured to decompose the first input signal into a first representation of the first input signal in a frequency domain, wherein the first representation comprises one or more elements representative of the first input signal; and wherein the decomposition module comprises one or more characterization blocks for characterizing the one or more elements of the first representation in the frequency domain.