id	author	title	date	pages	extension	mime	words	sentence	flesch	summary	cache	txt
fcis-31893	Zhu, Tianxiang	Research on Urban Sound Classification based on ConvNeXt-FECA Model	2025	7	.pdf	application/pdf	3829	190	33	Meanwhile, the choice of feature extraction method has a critical impact on model performance; among the single-feature methods, Spectrogram performs the best, while the combination of fused features, especially Spectrogram and MFCC, further enhances the model's classification ability. Performance metrics of the ConvNeXt-FECA model with different feature extraction methods.	cache/fcis-31893.pdf	txt/fcis-31893.txt