Table 9.

Validation results of unseen speakers with the multi speaker model.

Person	s1	s12	s13	s14	s16	s17	s18	s19	s20
Recog_rate	0.473	0.630	0.347	0.381	0.610	0.527	0.405	0.627	0.494
IQR	0.029	0.016	0.006	0.025	0.007	0.009	0.023	0.019	0.004