Effect of Concave Sound Reflecting Surfaces on Speech Intelligibility and Articulation Index.

Three different methods are used to assess speech intelligibility in spaces with and without concave sound reflecting surfaces: calculated articulation index (AI), measured rapid speech transmission index (RASTI), and modified rhyme tests (MRT) with occupants. Factors such as the room size, size of curvature, the speaker's position, and the background noise level are considered in the two on -site testing methods. The MRT results show unexpectedly significant deviation from results obtained through the other methods such that they are de-emphasized in all discussions. Results from rooms without curvatures show no significant differences between the AI and RASTI values; whereas, these differences are significant when rooms with curvatures are considered. A modification factor to be subtracted from calculated AI values to account for erosional effects of the curved surfaces is developed according to further analysis of the differences between the AI and RASTI values. The magnitude of the modification factors depends on all the above factors as well as the location of the listeners within the room. There are no clear indications of any dead spots, however, the sound foci from both the 2ft. and 8ft. curvatures have caused certain group locations to have smaller modification factors than that of all other locations. The magnitude of the developed modification factors ranges between 0.01, for the 16ft. curvature in the small rooms, to 0.17, for the 8ft. curvature in the large room with NC-45 and the speaker's position is on the center. This range is of almost the same magnitude as that of the erosional corrections to calculated AI due to elevated reverberation time. This range is also of almost same magnitude as that of improvement in calculated AI due to presence of visual cues.