Table 2.

Experimental parameters, including number of runs; set sizes; and minimum detection cost function (minDCF) parameters such as the cost of a false acceptance (CFA), cost of a false rejection (CFR), and prior probability (prior).

Experiment	Runs, n	Set size, total speakers				minDCF parameters
		Known	Unknown	Unknown only	Overlap	CFA	CFR	Prior
VoxCeleb: effect of search space size and known-overlap worst-case scenario	50	1000 to 7205 (varied known) 6000 (varied unknown)	163 (varied known) 150 to 1000 (varied unknown)	158 (varied known) 145 to 995 (varied unknown)	5	10	0.1	0.001
VoxCeleb: full-overlap worst-case scenario	20	1000 to 7205 (varied known) 6000 (varied unknown)	163 (varied known) 150 to 1000 (varied unknown)	0	163 (varied known) 150 to 1000 (varied unknown)	10	0.1	0.001
Mayo Clinic speech recordings: cross-task	20	500	55	50	5	1	1	0.01
Mayo Clinic speech recordings: within task	20	500a	55	50	5	1	1	0.01

^aWord repetition: 299 speakers; reading passage: 466 speakers.