ABSTRACT
In most criminal cases involving speaker recognition (SR) tasks an expert is supposed to define the possibilities and limitations of this forensic discipline. Quite often the expert will face questions by solicitors or attorneys such as: ‘Isn’t there a simple solution to this problem: after all, everyone seems capable of safely recognising people by their voices, even on the phone! And why can’t you use one of those computerised devices they have in detective films which can verify voices in just a couple of seconds?’ Quite obviously most individuals do not realise how difficult and complex the problem of forensic SR really is. So before elaborating on different methodological approaches it is possibly useful to clarify the fundamental differences between forensic and other SR applications. 1
