Evaluation of a Digits-in-Noise (DIN) test for auditory research and clinical application

Authors: Thu Ngan Dang1, Shangqiguo Wang2, Clément Gaultier3, Tobias Goehring4

1MRC Cognition and Brain Sciences Unit
2University of Hong Kong
3Institut Pasteur
4University of Cambridge

Background: Hearing speech in noise remains a major challenge for people with hearing loss. Digits-in-Noise (DIN) tests have been an accessible method to assess hearing ability and speech perception in noise. We developed a DIN test for auditory research and clinical audiology and evaluated its application for different listener groups including cochlear implant (CI) recipients.

Methods: Recordings of individual digits (0 to 9) from two native British English (male M and female F) were obtained in quiet and simulated noisy environments to elicit the Lombard effect. We first equalized for intelligibility (optimization) across digits using speech reception thresholds (SRTs) at 50% of single digits in noise. Then, we validated the test in-lab across listener groups (native – non-native, with or without CI simulation, typical hearing vs CI users), speaker voices, with or without Lombard effect using digit triplets in noise via an adaptive procedure. We also collected in-lab and online data to assess effects of digit optimization on DIN scores and slopes.

Results: The average SRT was in line with previous studies at -11.3 dB (-10 to -13 dB, Potgieter et al. 2016), with an average test slope of 15.3 %/dB (common range from 15 to 20 %/dB). The overall test measurement error was 0.8 dB. We found small effects on SRTs for English language background (0.6 dB, p<0.05), for the Lombard effect (0.4 dB, p<0.001) and sex of the speaker (0.4 dB, p<0.001). Small effects were found for digit optimization in-lab and online for SRTs (<1 dB) and test slopes. SRTs with CI simulations (-5.9 dB) were significantly (p<0.001) higher than without but lower than for CI listeners (-2.5 dB, ongoing). All found effects were smaller than the measurement error and the testing step size (2 dB).

Conclusion: Our evaluation of the DIN test suggests a robust measure of auditory perception and future development of DIN tests may benefit from a simpler procedure for more accessibility and inclusiveness.