Speaker adaptation with all-pass transforms | IEEE Conference Publication | IEEE Xplore