This study empirically evaluated the impact of deep neural network (DNN)-based noise reduction in hearing aids on the categorical perception of Mandarin Tones 1 and 2 for hard of hearing (HH) listeners in cafeteria noise at 0 and -5 dB signal-to-noise ratios (SNRs), compared to normal-hearing (NH) listeners and traditional hearing aid settings.
✦ The floor
Discussion
Signed responses from readers of the wire.
Audiologists fitting Mandarin-speaking hearing aid users should be aware that DNN-based noise reduction may alter categorical perception of tones; evaluation with tonal speech stimuli specific to the patient's language is advisable during fitting.
As AI-driven noise reduction becomes standard in hearing aids, understanding its unintended effects on tonal language perception is critical for fitting protocols serving the world's largest hearing-aid-using population.
- 01DNN-based noise reduction in hearing aids was evaluated for effects on Mandarin Tone 1 and Tone 2 perception.
- 02Categorical perception — the ability to reliably distinguish between tones — was the primary outcome.
- 03Hard-of-hearing Mandarin speakers were the study population.
- 04Study published in the American Journal of Audiology.
- 05Findings carry practical implications for hearing aid fitting in tonal-language communities.
Deep neural network-based noise reduction in hearing aids affects categorical perception of Mandarin Tones 1 and 2 in hard-of-hearing listeners.
studypartially supported- PMID
- 42525881
- DOI
- 10.1044/2026_AJA-25-00294.
- Journal
- American Journal of Audiology
- Publication type
- research_article
- Evidence level
- 2b
- Population
- Hard-of-hearing Mandarin-speaking listeners
- Intervention
- Deep neural network (DNN)-enhanced hearing aid noise reduction
- Comparator
- Hearing aid condition without DNN noise reduction (implied)
Primary outcomes
Categorical perception accuracy for Mandarin Tone 1; Categorical perception accuracy for Mandarin Tone 2