5th European Conference on Speech Communication and Technology

Rhodes, Greece
September 22-25, 1997

Estimation of Global Posteriors and Forward-Backward Training of Hybrid HMM/ANN Systems

J. Hennebert (5,2), Christophe Ris (1), Hervè Bourlard (3,2), Steve Renals (4), Nelson Morgan (2)

(1) TCTS, FPMs, Mons, Belgium (2) ICSI, Berkeley, CA, USA (3) IDIAP, Martigny, Switzerland (4) Computer Science, University of Sheffield, Sheffield, UK (5) CIRC, EPFL, Lausanne, Switzerland

The results of our research presented in this paper are two-fold. First, an estimation of global posteriors is formalized in the framework of hybrid HMM/ANN systems. It is shown that hybrid HMM/ANN systems, in which the ANN part estimates local posteriors, can be used to modelize global model posteriors. This formalization provides us with a clear theory in which both REMAP and "classical" Viterbi trained hybrid systems are uni_ed. Second, a new forward- backward training of hybrid HMM/ANN systems is derived from the previous formulation. Comparisons of performance between Viterbi and forward- back- ward hybrid systems are presented and discussed.

Full Paper

Bibliographic reference.  Hennebert, J. / Ris, Christophe / Bourlard, Hervè / Renals, Steve / Morgan, Nelson (1997): "Estimation of global posteriors and forward-backward training of hybrid HMM/ANN systems", In EUROSPEECH-1997, 1951-1954.