In my thesis work I measured different speech parameters in speech of alcohol intoxicated and sober people while they read a prescribed text. People can usually hear the difference between the speech of alcohol intoxicated and a sober person. The automatic recognition of intoxicated speech can help to decrease the accidents caused by alcohol intoxication. The effect of alcohol is influenced by lot of factors, which can affect the speech parameters also, such as mental state or fatigue. The effect of alcohol is affected by individual factors such as individual alcohol sensitivity or hunger. In my study I processed 29 voice files which I have got from the ELTE Institute of Phonetics. I preprocessed these files and I measured the following parameters: Jitter, Shimmer and harmonicity of sounds ‘a’, ‘i’, ‘u’, the speech tempo, articulation speed, rate of irregular voiced sounds, rate of stationary and transient sections in the whole speech.
I have found, that only the rate of the stationary and transient sections and the speech tempo changed tendentiously from the examined parameters.
I wish, the revealed relationships could increase the alcoholic speech detection