Research on Audio Recognition Based on the Deep Neural Network in Music Teaching

Table 3

Performance comparison of different methods for different types.

Model	Loss value	Training time	Type	mAP (%)	TP	FP	Precision	Recall	F1
CNN	0.0143	2 h 40 min	D1	97.5	77	0	0.98	0.97	0.98
			D2	98.81	83	0
			D3	99.92	62	1
			D4	98.74	76	2
GMM	0.0151	2 h	D1	97.53	78	0	0.98	0.96	0.97
			D2	100	83	0
			D3	99.85	61	3
			D4	98.01	75	3
R-CNN	0.0131	2 h 20 min	D1	97.50	77	0	0.98	0.97	0.97
			D2	98.81	83	0
			D3	99.92	59	0
			D4	97.75	72	5
SPP-YOLO-v4	0.0122	2 h 10 min	D1	97.51	78	0	0.99	0.99	0.99
			D2	98.82	83	0
			D3	99.90	62	1
			D4	98.94	79	3