gms | German Medical Science

Objective: Understanding speech well in acoustically complex listening environments with good sound quality is one of the main desires of hearing-aid users today. Modern signal processing techniques, such as spectral and spatial noise reduction (NR) algorithms allow for effective attenuation of noise, but may affect sound quality. The main objective of this study is to analyze the individual potential and actually achieved benefits due to NR in comparison to the hearing-aid users" preference ratings with respect to the strength of NR.

Methods: Seventy-two experienced hearing-aid users were fitted with Oticon More 1 hearing aids with either instant ear tips or individual earmolds using the NAL-NL2 fitting formula. After two weeks of accommodation, participants went through four field periods (FP). In each FP they were given two programs that differed in NR strength. After each FP, participants were asked for their preferred program, such that after the fourth FP the finally preferred NR strength on a 7-point scale was determined. In addition, situation-specific preference ratings were collected during the FPs. Speech-in-noise performance with preferred and initial settings was assessed before and after the FP. Furthermore, performance in the Audible Contrast Threshold (ACT™) test and closedness of acoustic coupling were measured.

Results: Speech-in-noise benefit greatly varied between participants, median improvements relative to a condition with no NR processing were 1.4 dB for mild and 4.1 dB for strong NR processing. This benefit was highly correlated with the closedness of acoustic coupling with ACT-score being the second-best predictor. Situation-specific preference ratings were clearer when the NR strength difference between the two programs was large and became less clear for smaller contrasts in NR strength. Final overall preferences for NR strength varied greatly between participants. However, the large spread of different final preferences was not related to possible achievable benefit, nor to other patient-specific factors, such as audiogram, ACT-score or closedness of acoustic coupling. Extreme NR settings (maximum or no NR) were dispreferred especially by poorer performers and those using closed fittings.

Conclusions: Large individual benefit in speech-in-noise performance due to NR strength is achievable, but not all hearing-aid users prefer such processing when asked to rate pairs of programs in the field. The results presented here may be used to improve the first fit of a hearing aid by means of individualized NR strength suggestions.