EN
The paper aims at presenting the results of an experiment checking the possibility of an accurate reconstruction of text sentence boundary done by humans and computer programme with possible application in presenting the output of automatic speech recognition systems. The results are compare with the assesment of received sentences acceptability.