During automatic speech processing a number of problems appear, and among them there are such as speech variation and different kinds of speech disfluences. In this article an algorithm for automatic detection of the most frequent of them (filled pauses and sound lengthenings) based on the analysis of their acoustical parameters is presented. The method of formant analysis was used to detect voiced hesitation phenomena and a method of band-filtering was used to detect unvoiced hesitation phenomena. For the experiments on filled pauses and lengthenings detection a specially collected corpus of spontaneous Russian map-task and appointment-task dialogs was used. The accuracy of voiced filled pauses and lengthening detection was 82%. And accuracy of detection of unvoiced fricative lengthening was 66%.