Hearing aid fitting and fine-tuning based on estimated individual traits

Area of Science:

• Audiology
• Signal Processing
• Human-Computer Interaction

Background:

• Hearing aid fitting is complex and often subjective.
• Current methods may not fully capture individual user preferences for noise reduction and distortion.

Purpose of the Study:

• To present a generalized concept for hearing aid fitting and fine-tuning based on estimated individual traits.
• To introduce a novel subjective assessment method, Combined Discrimination and Classification (CoDiCl), for capturing user preferences.
• To demonstrate the estimation of individual traits for a multidimensional noise reduction algorithm.

Main Methods:

• Utilized auditory model-based performance measures to generate algorithm candidates.
• Employed a multi-stimulus test (CoDiCl) for subjective user preference assessment.
• Reduced dimensionality using weighted combinations of Speech Intelligibility Index (SII) and Perceived Similarity Measure (PSM).

Main Results:

• Identified nine algorithm setting candidates for subjective rating.
• Differentiated between listeners based on preferences for noise suppression versus distortion.
• Demonstrated an exemplary estimation of individual traits for a coherence-based noise reduction algorithm.

Conclusions:

• Iterative improvement between subjective and objective assessments is key.
• Objective estimation of subjective traits may become possible with appropriate weightings.
• This approach can potentially lead to efficient fitting of modern hearing aid algorithms to individual users.