Table 3.

LOSO accuracies for each of the classifiers. The best-performing models for each feature type are red.

Features	Dim. Red. (n_comp)	LDA	DT	1NN	SVM	RF
LIWC	None	0.741	0.593	0.620	$\text{0.833}$	0.778
	LDA (1)	0.741	0.750	0.750	0.731	0.750
	PCA (20)	0.778	0.620	0.704	0.787	0.759
BERT	None	0.713	0.676	0.787	$\text{0.796}$	0.769
	LDA (1)	0.713	0.657	0.667	0.713	0.657
	PCA (2)	0.630	0.648	0.602	0.546	0.694
	PCA (20)	0.750	0.713	0.722	0.769	$\text{0.796}$
BERT + LIWC	None	0.750	0.657	0.667	$\text{0.824}$	0.806
	LDA (1)	0.750	0.731	0.731	0.741	0.731
	PCA (20)	$\text{0.824}$	0.620	0.657	$\text{0.824}$	0.796
BERT + CLAN	None	0.778	0.657	0.759	0.824	0.750
	LDA (1)	0.778	0.769	0.769	0.787	0.769
	PCA (20)	0.824	0.630	0.657	$\text{0.898}$	0.778
BERT + LIWC + CLAN	None	0.593	0.731	0.713	0.815	0.806
	LDA (1)	0.593	0.611	0.611	0.593	0.611
	PCA (20)	$\text{0.833}$	0.731	0.713	0.815	0.787
word vectors	None	0.759	0.731	0.694	0.259	0.694
	LDA (1)	0.759	0.741	0.731	0.759	0.759
	PCA (2)	0.676	0.620	0.565	0.259	0.620
	PCA (70)	$\text{0.796}$	0.648	0.759	$\text{0.796}$	0.787
i-vectors (VoxCeleb)	None	0.574	0.423	0.454	0.574	0.500
	LDA (1)	0.574	0.500	0.500	0.574	0.500
	PCA (2)	0.491	0.500	$\text{0.602}$	0.519	0.491
	PCA (10)	0.528	0.556	0.546	0.491	0.528
i-vectors (Pitt)	None	0.528	0.491	0.500	0.509	$\text{0.593}$
	LDA (1)	0.528	0.537	0.537	0.537	0.537
	PCA (2)	0.463	0.500	0.528	0.343	0.546
	PCA (20)	0.565	0.537	0.528	0.565	0.565
i-vectors (VoxCeleb + Pitt)	None	0.528	0.509	0.500	0.528	0.556
	LDA (1)	0.528	0.519	0.519	0.528	0.519
	PCA (20)	0.519	0.528	0.574	0.472	$\text{0.620}$
x-vectors (VoxCeleb)	None	0.583	0.620	0.509	0.546	0.574
	LDA (1)	0.583	0.593	0.593	0.583	0.593
	PCA (2)	0.472	0.537	0.491	0.454	0.491
	PCA (40)	$\text{0.639}$	0.583	0.528	$\text{0.639}$	0.583
x-vectors (Pitt)	None	$\text{0.546}$	$\text{0.546}$	0.472	0.528	0.481
	LDA (1)	$\text{0.546}$	0.500	0.500	0.537	0.500
	PCA (40)	0.537	0.481	0.435	0.528	0.491
x-vectors (VoxCeleb + Pitt)	None	0.639	0.602	0.519	0.620	0.509
	LDA (1)	0.639	0.509	0.509	0.630	0.509
	PCA (40)	$\text{0.657}$	0.574	0.546	0.593	0.593