In this paper, we investigated the parameter extraction to determine whether to drink or not by
comparing the voice characteristics before and after drinking through VTS. On the sea, rather than on the
road, there is a great deal of difficulty in pursuing drunken vessels by conducting drunk-tightening. The
characteristics of the drinker include characteristics such as pitch, formant, and sound size change. However,
it is necessary to find more exact feature parameters because it is very sensitive to environmental change
to determine before and after drinking. In order to observe the ship's operation using the VTS at sea, the
VHF radio carrier frequency band is used. In this case, speech is transmitted through voice compression,
and transmission and reception are performed using only frequency information of 300 to 3400 Hz
according to the characteristics of the vocoder. In order to judge whether or not you are drinking,
important information of less than 300Hz and 3400Hz is lost and a method is needed to cope with it.
Therefore, in this paper, we have studied the characteristics of voice signal before and after drinking using
distance measurement of effective frame, which is a time - domain drunk decision algorithm in VTS
environment.