This paper describes procedures to screening vocal diseases using acoustic speech data which is collected from ARS(Automatic Response System). Speech data from normal peoples and patients are collected and diagnosed. The collected materials are analyzed using several mathematical features such as jitter, shimmer etc. The classification is performed using artificial neural networks. Then the classification rate is compared with the case when voice data are recorded directly to the DAT(Digital Audio Tape). The performance of neural net discriminator was comparable to the previous work. As a result of this experiment, it is verified that ARS voice can be used screening pathological voice by acoustical signal on the telephone network.

Index Terms. Pathological voice; ARS; Neural Networks