. 2023 Jul 7;23(13):6212. doi: 10.3390/s23136212

Table 8.

CTENet prediction performance (in %) with and without multi-head attention Transformer.

Model Input	Database	Neural Architecture	Accuracy	Precision	F1 Score
MFCC Spectrum	RAVDESS	CTENet without MHAT	72.10	73.34	80.40
MFCC Spectrum	IEMOCAP	CTENet without MHAT	70.32	69.95	79.65
MFCC Spectrum	RAVDESS	CTENet with MHAT	78.00	78.75	84.37
MFCC Spectrum	IEMOCAP	CTENet with MHAT	79.00	74.80	82.20