 Hello and welcome to this pitch about estimating the distortion component of fearing impairment from attenuation-based model predictions using machine learning. First of all, what is the distortion component? In the right you can see the SRT, so the speech recognition threshold, which denotes the point where 50% of the spoken words are understood, depicted across the noise level. For normal hearing listeners and people who have some kind of attenuation component of hearing impairment, so increased absolute hearing thresholds, their hearing threshold defines their performance in rather or their SRT in rather calm environments. As the noise level increases, the noise level is defining their SRT and hearing impaired people with the A component can get as good as normal hearing listeners. People who also search for distortion components have more difficulties to understand speech and quiet, but they also have more difficulties to understand speech and noise environments. The A component directly correlates with the absolute hearing threshold, where there is no such thing for a D component. To determine these two parameters, one can do two SRT measurements, one in quiet, one in noise, and then fit these functions to it. However, this procedure is rather bothersome, so it would be nice to have some other approach to figure out this D component. What we did, we took SRTs, which were measured in a stationary noise, which was presented at 65 dB SPL from 315 different ears. Our assumption for modeling was that if we just use the absolute hearing threshold to adjust the A component for the simulations, then the model's prediction error will reflect, to some extent, the D component. We used three different models, namely the FrameMap auditory discrimination experiments on short fade, which is a machine learning-based approach, which can consider the A and the D component, although we just use the A component here, and the speech intelligibility index and a modified version of it, who both can just use the A component for making predictions. The results show that the D component seems to increase with a hearing loss, which is quite surprising. And if you're not thinking, well, why do you think that it works, I would like to invite you to join my presentation. And with that, I would like to thank you for your attention.