Speaker Recognition in Italy: evaluation of methods used in forensic cases

In this paper we present a specific corpus named ``Primula'', recorded and collected at the Phonetics Laboratory at the University of Calabria, reproducing characteristics and instruments usually to be found in legal cases. We then refer to this corpus in a first attempt to evaluate all the Forensic Speaker Recognition (FSR) methods used in Italy using a common data set. Preliminary results demonstrate, however, that much work has yet to be done in order to verify and validate the FSR methods especially when, as happens in Italy, the prosecutions' deductions and conclusions, and subsequently, the verdict, are primarily based on speaker identification.