Optimal Parameters for Measuring Multiband Auditory Brainstem Responses to Continuous Speech.

Every participant had normal hearing. The test is being built to characterise hearing loss in people who cannot tell us what they hear, and this study says nothing about how it performs in them. Participants were mostly in their early twenties and mostly women; the authors note a clear gender bias which could affect the absolute size and timing of the responses. Each experiment used a different group of people, so comparisons between experiments are comparisons between different participants. Test speed was estimated from signal-to-noise ratios rather than measured directly, and estimates for the slowest responders were extrapolated up to 120 minutes rather than recorded. The modelling assumed a flat, normal audiogram, and the modelled wave V timings did not perfectly match the measured ones. Everyone was tested at a single conversational level of about 65 dB SPL; the authors suggest a higher level might be faster but did not test it. The lowered-pitch recordings were re-synthesised from the original narrators' audio rather than newly recorded voices. Recording sessions were held to about two hours because people became restless and recordings got noisy beyond that.

Funded by a Hearing Health Foundation Emerging Research Grant and the University of Minnesota's Undergraduate Research Opportunities Program. The authors declared no competing interests.

Do not take this as evidence the test is ready for the clinic or that it will work in people with hearing loss. All 55 participants had normal hearing and were mostly in their twenties, and the reported test times are estimated from signal-to-noise ratios, with the slowest ones extrapolated rather than measured.