Compensating for Mismatch in High-Level Speaker Recognition | IEEE Conference Publication | IEEE Xplore