Recently, feature representation by learning algorithms has drawn great attention. In the music domain, it is either unsupervised or supervised by semantic labels such as music genre. However, finding discriminative features in an unsupervised way is challenging, and supervised feature learning using semantic labels may involve noisy or expensive annotation. In this paper, we present a feature learning approach that utilizes artist labels attached in every single music track as an objective meta data. To this end, we train a deep convolutional neural network to classify audio tracks into a large number of artists. We regard it as a general feature extractor and apply it to artist recognition, genre classification and music auto-tagging in transfer learning settings. The results show that the proposed approach outperforms or is comparable to previous state-of-the-art methods, indicating that the proposed approach effectively captures general music audio features.
View on arXiv