A Determination of Drinking According to Types of Sentence using Speech Signals

In this paper, we investigated the accuracy of negative utterance sentences in order to judge drinking. Drinking makes it difficult to pronounce specific sounds correctly. A number of speech parameters have been studied to identify inaccurate pronunciation features. Generally, the method of grasping the inaccurate pronunciation is analyzed using the change of pitch and formant. Formants have various information such as the individual's uniqueness, clarity, and physical condition, and pronunciation status can also be analyzed. Analysis of the pronunciation through the pronunciation agency may cause difficulty in the analysis due to the pronunciation difference of the individual. On the other hand, analysis of pronunciation through the vocal organs is easy to analyze because consonants or vowels are applied only to the consonant vowel rather than to the pronunciation. Especially when drinking alcohol, the vocal cord is dehydrated due to alcohol. Therefore, it is difficult to elaborate voices or to utter certain sounds. Therefore, in this paper, we have studied the possibility of extracting the alcohol discrimination parameter with the change of the accuracy of the voiced sound according to the vocal sentence vowels. As a result, the longer the sentence, the more pronounced error was obtained and the meaningful result was obtained when the number sentence was spoken.