Do Humans and speaker verification system use the same information to differentiate voices?

摘要

The aim of this paper is to analyze the pairwise comparisonsof voices by a speaker verification system (ALIZE/Spk) andby human. A database of familial groups of 24 speakers wascreated. A single sentence was chosen for the perception test.The same sentence was used the test signal for theALIZE/Spk trained on another part of the corpus. Resultsshows that the voice proximities within a familial group werewell recovered in the speaker representation by ALIZE andmuch less returned in the representation from perception test.