Hierarchical disentangled representation learning for singing voice conversion | IEEE Conference Publication | IEEE Xplore