Harmonic Classification with Enhancing Music Using Deep Learning Techniques

Table 4

Results of various training configurations of proposed model systems.

Text sets	Method	Train set	Weighted	Correct	Fifth	Relative	Parallel	Other
GS	CK1	GS^MTG	75.3	68.2	6.8	7.1	4.3	14.1
	CK2	BB^TV	57.6	47.5	6.7	12.8	16.8	17.7
	CK3	GS^MTG and BB^TV	69.5	61.6	6.9	8.7	6.5	16.6
	EDM^A		65.9	57.4	7.6	6.8	11.0	17.8
	EDM^M		70.4	63.5	8.8	2.6	6.5	18.7
	EDM^T		44.9	33.9	8.7	15.7	9.7	32.5
	QM		50.8	39.8	12.0	13.5	4.9	31.3
BB^TE	CK1	GS^MTG	72.9	62.8	7.8	13.4	12.7	4.4
	CK2	BB^TV	84.0	77.4	9.2	5.1	4.5	5.0
	CK3	GS^MTG and BB^TV	80.0	71.0	9.9	9.3	6.6	4.2
	EDM^A		78.9	70.8	11.6	3.0	5.8	9.3
	EDM^M		30.0	14.8	2.4	16.3	42.2	25.2
	EDM^T		75.8	66.9	12.7	6.5	2.9	12.0
	QM		61.0	52.3	11.9	4.4	8.5	23.9