P67 An open generic specification for the Digits-in-Noise test
Smits and colleagues [2013, JASA 133(3):1693] developed the Digits-in-Noise (DIN) test, which requires individuals to repeat three spoken numbers (a digit triplet) presented in noise, for clinical usage. They determined the criterion validity of the DIN test by comparing results on the test with results on the sentence speech-recognition threshold (SRT) from Plomp and Mimpen [1979, Audiol 18:43]. Both tests strongly correlated with pure tone audiometry (PTA) averages. While PTAs require strictly calibrated apparatus, speech tests remain informative over a range of presentation levels. Moreover, in comparison to the sentence SRTs, the DIN test only requires understanding of short, closed set tokens. Therefore, its application is less restricted by severity of hearing impairment or linguistic skills, making it applicable to, cochlear implant users and children. Over the past years, the DIN test has grown to be the number one tool for easy assessment of hearing acuity both in the clinic and online (hoortest.nl). It has even been translated to English and promoted by the World Health Organization for mobile audiometric assessment (hearWHO).
While first validated as a screening tool for detecting incipient hearing loss, nowadays the DIN is also used to assess hearing aid and cochlear implant fitting, which resulted in a vast amount of normative data for normal hearing and different hearing-impaired populations across multiple languages. However, to accommodate the specific needs of these diverse populations, purposes and situations, slightly different variations of the DIN test have been used that differ, for example, in the way speech is presented (diotic or antiphasic), or in the starting signal-to-noise ratio of the staircase procedure.
Similarly, for our use across various projects, we have created different versions of the DIN test: in MATLAB, online, and in Robot Operating System (ROS, https://www.ros.org), using the original Dutch DIN material (Smits et al., 2013) that has been made available to us (Cas Smits from Amsterdam UMC). To allow for compatible implementations and comparable results, we established a specification defining the test options and results. Each project implementing a DIN test following this specification will be able to run the known variations of the DIN test observed in the literature and in the clinic; therefore, defining the results data structure, and simplifying data exchange across projects, across centers, or even across languages.
The specification is published with an open CC-BY license on the dB SPL GitHub repository at https://github.com/dbSPLab/din-specification. We invite comments from expert users of the DIN to make the specification evolve.