15th Speech in Noise Workshop, 11-12 January 2024, Potsdam, Germany 15th Speech in Noise Workshop, 11-12 January 2024, Potsdam, Germany

P60Session 2 (Friday 12 January 2024, 09:00-11:30)
Development of the Turkish Digits-in-Noise test

Soner Türüdü, Thomas Koelewijn, Deniz Başkent
Department of Otorhinolaryngology/Head and Neck Surgery, University Medical Center Groningen, University of Groningen, Netherlands

Background: The Digits-in-Noise (DIN) test is a closed-set, adaptive, simplified Speech-in-Noise (SIN) test that uses digit triplets in a staircase procedure to measure Speech Reception Threshold (SRT) in noise. DIN was originally developed in Dutch and was later adapted to other languages. This study aims to create new Turkish DIN materials and assess the test-retest reliability of the Turkish DIN test. Further, we investigate the effects of speaker variability and diotic and dichotic sound presentation methods as additional measures for reliability, as at least the latter was clearly shown to affect the DIN outcomes of previous materials.

Methods: We recorded Turkish digits (0-9) from two male and two female native speakers, from which the ten clearest instances for each digit were selected. The selected recordings were then high-pass filtered at 80 Hz to remove low-frequency noise, trimmed, and ramped for onset and offset. In conducting the Turkish DIN test, contrary to fixed triplets used in some DIN tests, our digit triplets are not predetermined but randomly selected from a pool for each digit, ensuring no three digits are identical in any given trial. Each triplet is consistently composed of recordings from the same speaker, with 150 ms silence between digits and 500 ms silence at both ends. For testing, these digit triplets are being presented in a fixed-noise condition, configured to have the same long-term average spectrum as the digits spoken by each speaker. The presentation level is set at 65 dB SPL. The test starts with a -16 dB starting Signal-to-Noise Ratio (SNR), with an initial 4 dB step size, which is adjusted to 2 dB after the first correct response. A 4x2x2 factorial design has been applied to assess the impact of speaker variability (2 male, 2 female) and sound presentation methods (diotic, dichotic). Block randomization has been used to distribute the eight test conditions across two sessions - an initial test and a retest after a 15-minute break - for 15 participants to evaluate the stimuli's effectiveness and the test's reliability through repeated measures.

Results and Discussion: The data collection is ongoing and is expected to be completed by the 15th Speech in Noise (SpiN) Workshop. Upon completion of the data analysis, this study aims to determine the most suitable speaker’s recordings for refining the Turkish Digits-in-Noise test in future studies.

Last modified 2023-11-29 15:13:43