Unsupervised Interpretable Representation Learning for Singing Voice Separation | IEEE Conference Publication | IEEE Xplore