A new gain model for speech enhancement | IEEE Conference Publication | IEEE Xplore