A hierarchical approach for speech-instrumental-song classification

SpringerPlus

Table 2 Accuracy (in %) of speech, music classification

Classification scheme	Experiment setup	Type of signals		Featutre set
			ZCR, STE based features	Δ- Δ²	Proposed audio texture
K-means	T1	Speech	50.55	60.00	73.89
		Music	74.07	74.07	85.93
		Overall	64.67	68.08	81.11
	T2	Speech	48.15	57.52	71.85
		Music	71.04	72.10	84.40
		Overall	61.87	65.91	79.38
MLP	T1	Speech	71.11	78.50	78.33
		Music	90.37	75.92	88.15
		Overall	82.67	77.02	84.22
	T2	Speech	68.52	74.92	74.07
		Music	86.63	72.84	84.40
		Overall	79.38	73.72	80.27
SVM	T1	Speech	73.89	86.00	78.33
		Music	90.74	81.48	89.26
		Overall	84.00	83.40	84.89
	T2	Speech	69.63	83.28	76.67
		Music	88.86	79.75	85.40
		Overall	81.16	81.25	81.90
RANSAC	T1	Speech	75.00	88.00	96.11
		Music	92.96	85.93	97.78
		Overall	85.78	86.80	97.11
	T2	Speech	74.07	85.62	93.70
		Music	90.59	82.47	93.81
		Overall	83.98	83.81	93.77