No abstract

No abstract

This paper summarizes the main results from the Speaker Veri cation (SV) research pursued so far in the CAVE project. Di erent state-of-the art SV algorithms were implemented in a common HMM framework and compared on two databases : YOHO (ofce environment speech) and SESP (telephone speech). This paper is concerned with the di erent design issues for LR-HMM-based SV algorithms which emerged from our investigations and which led to our current SV system, which delivers Equal Error Rates below 0 . 5 % o n a v ery realistic telephone speech database.

IRISA(1) KTH(2) KUN(3) ENST(4) UBS-Ubilab(5) IDIAP(6)

No abstract