Speech Trajectory Discrimination Using the Minimum Classi cationError

In this paper, we extend the Maximum Likelihood (ML) training algorithm to the Minimum Classiica-tion Error (MCE) training algorithm for discriminatively estimating the state-dependent polynomial coeecients in the stochastic trajectory model or the trended HMM originally proposed in 2]. The main motivation of this extension is the new model space for smoothness-constrained, state-bound speech trajectories associated with the trended HMM, contrasting the conventional, stationary-state HMM which describes only the piecewise-constant \degraded trajectories" in the observation data. The discriminative training implemented for the trended HMM has the potential to utilize this new, constrained model space, thereby providing stronger power to disambiguate the observational trajecto-ries generated from nonstationary sources corresponding to diierent speech classes. Phonetic classi-cation results are reported which demonstrate consistent performance improvements with use of the MCE-trained trended HMM both over the regular ML-trained trended HMM and over the MCE-trained stationary-state HMM. Permission to publish this abstract separately is granted.