ESCA Workshop on Automatic Speaker Recognition, Identification, and Verification
We present the project of an Italian speech database collation over the Public Switched Telephone Network. The goal of this database is to create a reference material for speaker recognition assessment, as regard the Italian language. It is part of a wider Italian national project managed by the ISPT (Istituto Superiore delle Poste e Telecomunicazioni) that is the tutor of the "Italian Association of Acoustic Databases" (AIDA). After the production of a set of 6CD for speech recognition assessment and the design of two databases for speech synthesis, the AIDA4 corpus, for the assessment of speaker recognition over the telephone line, has been designed. In this paper we describe the contents of this database, the collation modalities and the technical set up of the collation. As far as possible in designing this database we consider the different kind of speaker recognition systems (text-dependent, text-independent, listening tests, etc.). The start of the collation is planned for the second half of this year, and it will take about eight months. A total of 500 speakers will take part to the realisation of this database. The database will be ownership of the ISPT, that is also the official distributor of all AIDA corpora.
Bibliographic reference. Carlo, Andrea Di / Falcone, Mauro / Paoloni, Andrea (1994): "Corpus design for speaker recognition assessment", In ASRIV-1994, 47-50.