Accurate Prediction of Speech Intelligibility without the Use of In-Room Measurements

The speech transmission index (STI) has been shown to be an accurate predictor of speech intelligibility in rooms, and the computationally efficient rapid STI (RASTI) method has recently become an International Electrotechnical Commission (IEC) standard. Instruments exist to measure the STI after a sound system has been installed and is operating, but until now the STI method has not been implemented and its accuracy verified in a sound system modeling program. Such an implementation has a fundamental advantage in that it does not require acoustic measurements from the room as input; this means that intelligibility can be prdcicted in unbuilt or inaccessible rooms solely on the basis of modeled rather than actual acoustic behavior. A new microcomputer based implementation of the STI method is described along with the results of an experiment designed ot test its accuracy. The accuracy of the new method is shown to be essentially the same as the accuracy of predictions based on in-room measurements. These results show that speech intelligibility can be predicted reliably without using acoustic measurements.