Table 3.

Details of music emotion recognition algorithms based traditional method.

Method	Dataset	Features	Classifier/regressor	Performance
Avramidis et al. (2021)	DEAP	PSD, HFD, MFD, MADFA	RBF-SVM	Accuracy of 67% in Binary classification of arousal.
Hasanzadeh et al. (2021)	15 recruited listened to 7 songs	Spectrograms from Morlet wavelet transform	Fuzzy Parallel Cascades	2 types regression of Valence with RMSE 0.089.
Thammasan et al. (2016c)	15 recruited listened to 16 songs selected from MIDI	HFD	SVM	3% performance increase over the non-filtered.
Zainab and Majid (2021)	27 recruited listened to bilingual audio music of five genres	PSD, HFD, Hjorth Parameters. A series of linear measures of time domain	Hyper Pipes	Accuracy of 83.95% in quaternary classification..
Thammasan et al. (2016b)	12 recruited listened to 16 songs selected from MIDI	FD for EEG and handcraft feature for music	SVM	MCC of 84.17 and 90.25% in binary classification of arousal and valence, respectively.
Naser and Saha (2021)	DEAP	Wavelet transform, functional connectivity, graph-theory based features	RBF-SVM	Accuracy of arousal, valence, and dominance were 22.50, 14.87, and 19.44% above the empirical chance-level, respectively.
Thammasan et al. (2017b)	DEAP	PSD, HFD	Kernel SVM, MLP, Decision Tree	An average of 5% classification improvement of Unfamiliar set above familiar set in three methods.
Shahabi and Moghimi (2016)	19 recruited listened to six classical music excerpts	Connectivity matrices	SVM	Joyful vs. neutral, joyful vs. melancholic and familiar vs. unfamiliar trials reach accuracy of 93.7, 80.43, and 83.04%, respectively.
Lin et al. (2009)	26 recruited	PSD	One-against-one scheme SVM	Accuracy of 92.57% in quaternary classification.
Bhatti et al. (2016)	30 recruited listened to 4 genres of music	Latency to Amplitude Ratio, PSD, Wavelet transform	MLP, KNN, SVM	Accuracy of 78.11% (MLP) in quaternary classification.