BUT VOiCES 2019 System Description
This is a description of our effort in VOiCES 2019 Speaker Recognition challenge. All systems in the fixed condition are based on the x-vector paradigm with different features and DNN topologies. The single best system reaches 1.2 relative improvement. The open condition allowed us to use external data which we did for the PLDA adaptation and achieved less than 10 improvement. In the submission to open condition, we used 3 x-vector systems and also one i-vector based system.
READ FULL TEXT