A linked-HMM model for robust voicing and speech detection | IEEE Conference Publication | IEEE Xplore