A new speech corpus in Spanish for speaker verification

In this paper we present a new database with speech recordings in Spanish. The database contains recordings of 54 native Spanish speakers. It is appropriate to be used in the development and testing of better Speaker Verification systems. The recording procedure, equipments and speech tasks are detailed. Experiments using the GMM-UBM speaker verification methodology were performed. The methodology of these experiments is described. Text-independent and text-dependent models were tested. The best results were obtained with text-independent models, EER = 3.04%. These kind of models also offer other advantages over text-independent models. According to the results this database is appropriate for speaker verification and identification research. Anyone interested in using this database for research purposes is welcome to contact the authors.