Dynamic Convolution With Global-Local Information for Session-Invariant Speaker Representation Learning | IEEE Journals & Magazine | IEEE Xplore