Speech intelligibility in real and virtual classrooms

Abstract

In current research by the authors, auralization is being used to identify the optimal reverberation times and signal-to-noise level differences for speech in classrooms. The work presented here was initiated to validate the auralization procedure in comparison with live listening tests in real classrooms. Reverberation and noise conditions were created in both real and virtual classrooms, and speech intelligibility tests performed. Two architecturally identical classrooms which have different amounts of sound absorption were selected for real classroom speech intelligibility tests, and their acoustical parameters were measured. Speech and noise sources were placed at the fronts and backs of the classrooms. Speech intelligibility tests were performed at three positions with normal-hearing subjects. The modified rhyme test (MRT) and the noise babble were generated by loudspeakers. The classrooms and speech intelligibility test conditions were then simulated for auralization using CATT-Acoustics, based on the real classroom measurement results. Virtual speech intelligibility tests were performed on the same subjects in the laboratory using headphones. The MRT results in the real and virtual classrooms were compared.