Passato Presente e Futuro nelle Analisi di Speaker Recognition

In a scientific field there is an increasing need for a protocol in speaker recognition analysis. This need responds fully to a request of warranty and protection of all the parts involved in a criminal lawsuit. The current contribution doesn’t have the purpose to work out such a problem or to impose a precise methodology, but it has the only aim to point out the speaker recognition’s state in court rooms and in university’s research laboratories. Speaker recognition for forensic purpose is currently carried out both with subjective and objective methodologies. The subjective ones that will be presented are the auditory-perceptive, the spectrographic and a sort of linguistic-dialectological method; concerning the objective ones we will report the parametric, the semiautomatic and the combined method. Besides a series of basic steps together with arising problems in a vocal signal examination are presented as follows: signal acquisition, identification, extrapolation and measure of the parameters, use of specific equipments and algorithms, statistic comparison and individuation of a suitable reference community. These steps could be a basis and a starting point for a future protocol to which should follow the establishment of a professional register for forensic phonetics experts.