Development of speech recognition in noise and reverberation test with multichannel auralizations of real rooms

In recent years, a major step forward in improving the accuracy of hearing diagnostics and quality control of hearing rehabilitation in Finland has been the development and implementation of Finnish Matrix Sentence Test (FMST) [2]. However, stimuli in this and other similar tests, that measure speech intelligibility in noise, are commonly presented with little to no reverberation, although it is clear that the difficulties in understanding speech in every day life often arise from the reverberation in the surrounding space. The overall aim of this work is to investigate how advanced spatial sound technologies can be used to expand the current methodology and to develop novel tools for hear- ing diagnostics with more complex and authentic sound scenes. In the current study, we have made multihannel auralizations of two real spaces and compare the measured speech reception thresholds with these auralizations to the original FMST test results with normal hearing and hearing impaired listeners.