Speech-guided source separation using a pitch-adaptive guide signal model | IEEE Conference Publication | IEEE Xplore