Recent work has shown that the end-to-end approach using convolutional neuralnetwork (CNN) is effective in various types of machine learning tasks. Foraudio signals, the approach takes raw waveforms as input using an 1-Dconvolution layer. In this paper, we improve the 1-D CNN architecture for musicauto-tagging by adopting building blocks from state-of-the-art imageclassification models, ResNets and SENets, and adding multi-level featureaggregation to it. We compare different combinations of the modules in buildingCNN architectures. The results show that they achieve significant improvementsover previous state-of-the-art models on the MagnaTagATune dataset andcomparable results on Million Song Dataset. Furthermore, we analyze andvisualize our model to show how the 1-D CNN operates.
展开▼
