SESSION 2: DSP IN RECORDING Fast Monaural Separation of Speech Niels Henrik Pontoppidan 1 and Mads Dyrholm 2 1 Technical University of Denmark, Lyngby, Denmark 2 Copenhagen University Hospital, Copenhagen, Denmark We have investigated the possibility of separating signals from a single mixture of sources. This problem is termed the Monaural Separation Problem. Lars Kai Hansen has argued that this problem is topologically tougher than problems with multiple recordings. Roweis has shown that inference from a Factorial Hidden Markov Model, with nonstationary assumptions on the source autocorrelations modeled through the Factorial Hidden Markov Model, leads to separation in the monaural case. By extending Hansens work we find that Roweis� assumptions are necessary for monaural speech separation. Furthermore we develop a hierarchical Viterbi algorithm for the Factorial Hidden Markov Model yielding a significant decrease in complexity of inference.