Emotional Video to Audio Transformation Using Deep Recurrent Neural Networks and a Neuro-Fuzzy System

Table 6

Human evaluation with 10 samples generated with the DEAP dataset using the current model.

	Target MOS		Obtained MOS
	Valence	Arousal	Valence	Arousal
1	5.39	5.32	5.40 ± 2.80	5.20 ± 2.91
2	5.58	5.18	4.10 ± 2.47	4.80 ± 2.74
3	5.02	5.29	5.00 ± 2.21	5.70 ± 2.50
4	5.53	5.55	6.00 ± 2.26	5.60 ± 2.84
5	5.01	5.44	4.60 ± 2.37	4.90 ± 2.69
6	5.49	4.20	4.40 ± 2.63	4.90 ± 2.81
7	5.30	5.04	5.40 ± 2.99	5.10 ± 3.18
8	4.72	5.50	4.70 ± 2.50	5.50 ± 2.51
9	5.13	5.21	3.60 ± 2.32	4.30 ± 3.16
10	5.11	4.92	3.80 ± 2.74	4.50 ± 3.27