Data-Driven Decision-Support System for Speaker Identification Using E-Vector System

Table 6

Comparison of Avg. micro-F1and Std. Dev. micro-F1 SD of multiple-utterance identification.

Micro-F1	Different models
	GMM			HMM
	E-vector	MFCC	HDCC	E-vector	MFCC	HDCC
Avg.	↑0.982	0.978	0.862	↑0.935	0.924	0.879
Std. Dev.	↓0.018	0.023	0.139	↓0.039	↓0.039	0.090
	Different corpus databases
	TIMIT			LibriSpeech
	E-vector	MFCC	HDCC	E-vector	MFCC	HDCC
Avg.	↑0.936	0.930	0.771	↑0.982	0.972	0.971
Std. Dev.	↓0.035	0.037	0.071	↓0.025	0.034	0.029