A speech database was collected over the Italian Public Switched Telephone Network from more than 2000 telephone customers. An automatic workstation has been developed in order to perform the collection; it was connected to two speech recognizers that have been tested on-line during the database collection. Results and solutions to different problems faced during the collection are reported.
Bibliographic reference. Castagneri, G. / Fabbrizio, G. Di / Massone, A. / Oreglia, M. (1993): "SIRVA - a large speech database collected on the Italian telephone network", In EUROSPEECH'93, 199-201.