Hybrid Y-Net Architecture for Singing Voice Separation | IEEE Conference Publication | IEEE Xplore