14th Speech in Noise Workshop, 12-13 January 2023, Split, Croatia 14th Speech in Noise Workshop, 12-13 January 2023, Split, Croatia

P29Session 1 (Thursday 12 January 2023, 15:30-17:30)
Comparing Digits-in-Noise test implementations on various platforms with normal hearing individuals

Soner Türüdü, Thomas Koelewijn
Department of Otorhinolaryngology/Head and Neck Surgery, University Medical Center Groningen, University of Groningen, Groningen, Netherlands
Research School of Behavioral and Cognitive Neuroscience, Graduate School of Medical Sciences, University of Groningen, Groningen, Netherlands

Etienne Gaudrain
Lyon Neuroscience Research Center, CNRS UMR5292, Inserm U1028, UCBL, UJM, Lyon, France
Department of Otorhinolaryngology/Head and Neck Surgery, University Medical Center Groningen, University of Groningen, Groningen, Netherlands

Deniz Başkent
Department of Otorhinolaryngology/Head and Neck Surgery, University Medical Center Groningen, University of Groningen, Groningen, Netherlands
Research School of Behavioral and Cognitive Neuroscience, Graduate School of Medical Sciences, University of Groningen, Groningen, Netherlands

Background: The Digits-in-Noise (DIN) test measures speech reception threshold in noise using digit triplets in a staircase procedure and is commonly used as a screening tool in large populations. This has resulted in an abundance of normative data for a variety of populations in different languages. Previous studies used modified versions of the DIN-test to meet the needs of these populations. We aimed to evaluate these variations' effect on DIN scores on varying platforms in young normal-hearing adults.

Methods: We have created different versions of the DIN test in MATLAB and online using the original Dutch DIN materials (Smits et al., 2013). Fifteen and eight normal-hearing individuals participated in the lab and online sessions, respectively. DIN-test implementations had three variables: (1) Mixing Method determined whether the presentation level of speech, noise, or the mix of both (not implemented online) was fixed at 65 dB SPL; (2) Starting signal-to-noise ratio (SNR) could be High (0 dB) or Low (-16 dB); (3) Finally, Sound Presentation could be diotic or dichotic (speech phase reversed in one ear).

Results: A three-way repeated measures ANOVA showed that the DIN thresholds for dichotic compared to diotic Sound Presentation were significantly lower in MATLAB (SNR= -16.4 dB vs -10.1 dB, respectively) and online (SNR= -15.1 dB vs -9.5 dB, respectively) platforms. The MATLAB results also showed a significant main effect for Mixing Methods (SNR for fixSp = -12.9 dB, fixNse = -13.4 dB, fixMix = -13.5 dB), Starting SNR (High = -12.9 dB, Low = -13.6 dB), and a significant three-way interaction in MATLAB version. Online there was no significant effect for Mixing methods (SNR for fixSp = -12 dB, fixNse = -12.6 dB), and starting SNR (High = -12.2 dB, Low = -12.5). Post-hoc analysis on the MATLAB data revealed the Mixing Method, Starting SNR, and their interaction to be significant only in the dichotic conditions.

Conclusions: Dichotic listening resulting in lower DIN scores may be due to speech phase reversal in one ear, which helps separate speech from noise. DIN measurements should ideally be immune to changes in Mixing Method and Starting SNR. We observed a significant effect of these variables only in the dichotic condition of the MATLAB version; however, the effects were small and limited to a maximum of 1.3 dB SNR, and it appears that significance was only achieved because the participant group was homogeneous, young, and had normal hearing.

Last modified 2023-01-06 23:41:06