Sex, age, emotional state of the subject, pathological voices... are characteristics that allow us recognise different speakers. The accent difference persons of different regions, indeed their tone establish inequality between analogous voices; intensity and timbre differ voices with similar tones; pauses, tempo, idea's speed... condition an unique and discriminate voice for each speaker. Previous studies have researched the possibility of studying the speaker idiosyncrasies on phonetic regularities, by mean of temporal parameters like the segmental duration. This kind of characteristics have the property of a very easy measurement, they are very stables by the time and have a rriinimal change in function of the different environment. All this characteristics allow us, in a practical term, remark the idiosyncrasic pattern of each speaker.