Background: Hearing aids use various signal processing settings, such as linear and non-linear amplification and noise reduction algorithms, to enhance speech recognition in noise. Speech reception thresholds (SRT) indicate the minimum signal-to-noise ratio needed for speech recognition, tested at different task demands (50% and 80% accuracy). Noise types like speech-shaped noise (SSN) and four-talker babble present different challenges due to their distinct acoustic properties. Speech-state noise is continuous and predictable and has been suggested to be more straightforward to manage with linear amplification. At the same time, the dynamic four-talker babble benefits more from adaptive algorithms such as non-linear amplification and noise reduction. We aimed to examine whether the type of noise and task demand moderates signal processing settings.

Rationale: Acquiring knowledge of how different signal processing settings affect speech recognition in various noise environments, particularly cognitive load at different task demands, can be used to advance hearing aid technology and improve user experience.

Methods: Data from 215 individuals wearing hearing aids (mean age = 60.8 years, SD = 8.8) from the n200 database were used. The Hagerman task was used to assess speech recognition in noise performance. The main and interacting effects of hearing aid signal processing settings (linear amplification, non-linear amplification, and noise reduction), noise type (SSN vs. four-talker babble), and task demand (50% vs. 80% accuracy) on speech recognition performance were tested. SNRs were estimated for each signal processing setting at both 50% and 80% accuracy across the two noise conditions: four-talker babble and SSN. Linear mixed effects modeling was employed to analyze the data.

Results: Preliminary results suggest that the interaction between task demand and noise type influenced the effect of signal processing settings. Noise reduction was more effective in 4-talker noise compared to SSN noise. At higher task demands (80% SRT), noise reduction differed from nonlinear fast compression and linear amplification, which also showed differences from each other. At lower task demands (50% SRT), nonlinear and linear amplification showed slight differences, but noise reduction remained the most effective approach in both scenarios.

Conclusions: Noise reduction showed strong effectiveness in four-talker babble, especially under higher task demands (80% SRT). In contrast, at lower task demands (50% SRT), non-linear and linear amplification performed similarly in SSN. These results show the benefits of tailoring hearing aid settings to complex noise conditions, but that further research is needed to refine hearing aid technology for real-world listening environments.