This research looks into parameters for reducing errors of judgment intoxication by comparing speech characteristics before and after drinking using speech analysis. In this study a drinking is tested to speech signal under overloading condition, there are many of some errors in unstable conditions. Therefore, it should be determined by using judgment algorithm that is not affected by overloading. So we have searched valid frames using energy of speech before and after drinking were extracted and drinking possibility was identified by using deviation from the distance of frames. In terms of speech frame energy its value changes greatly in overloaded but energy ratio largely stays the same in normal. Therefore, after applying overloaded signals to proposed method, we get this similar result with a normal input signal was obtained