Effect of digital noise reduction processing on subcortical speech encoding and relationship to behavioral outcomes
Subong Kim1, Mary Schroeder2, Hari M Bharadwaj3
1Department of Communication Sciences and Disorders, Montclair State University, Montclair, NJ, USA.
Scientific Reports
|July 2, 2025
Summary
Individual differences in speech noise reduction (NR) benefits are not well understood. This study uses brainstem responses to objectively measure how the brain processes NR speech, linking these measures to behavioral outcomes.
Area of Science:
- Neuroscience
- Audiology
- Speech Processing
Background:
- Perceptual benefits of digital noise reduction (NR) in speech vary significantly across individuals.
- The underlying physiological mechanisms for these individual differences, particularly concerning noise tolerance and sensitivity to NR-induced distortions, remain understudied.
- Objective measures are needed to understand the neural basis of individual responses to processed speech.
Purpose of the Study:
- To develop objective measures of speech encoding in the ascending auditory pathway.
- To investigate if these measures can predict individual noise tolerance and sensitivity to NR-processed speech.
- To explore the relationship between subcortical speech encoding and behavioral outcomes with NR.
Main Methods:
- Utilized speech-evoked brainstem responses (e.g., to the syllable /da/) as objective measures.
- Assessed sensitivity to noise and NR processing using response-to-response correlation metrics and spectral amplitude ratios.
- Quantified NR effects on consonant and vowel portions of the brainstem response.
- Analyzed frequency-following response (FFR) f0 amplitude ratios and correlated them with behavioral accuracy.
Main Results:
- Speech-evoked brainstem responses demonstrated sensitivity to both the addition of noise and NR processing.
- NR effects on consonant and vowel encoding were reliably quantified.
- Frequency-following response (FFR) f0 amplitude ratios correlated significantly with behavioral accuracy in NR conditions.
- These findings indicate the feasibility of using brainstem measures to study NR effects on speech encoding.
Conclusions:
- Investigating NR effects on bottom-up speech encoding using brainstem measures is feasible.
- Individual subcortical encoding patterns of NR-processed speech may predict behavioral outcomes.
- This approach offers a potential pathway to understand and personalize speech processing technologies.
