Speaker detection without models | IEEE Conference Publication | IEEE Xplore