A Constrained Baum-Welch Algorithm for Improved Phoneme Segmentation and Efﬁcient TrainingComputer Science Department
Date of Original Version1-1-2006
Abstract or DescriptionWe describe an extension to the Baum-Welch algorithm for training Hidden Markov Models that uses explicit phoneme segmentation to constrain the forward and backward lattice. The HMMs trained with this algorithm can be shown to improve the accuracy of automatic phoneme segmentation. In addition, this algorithm is signiﬁcantly more computationally efﬁcient than the full BaumWelch algorithm, while producing models that achieve equivalent accuracy on a standard phoneme recognition task.
Citation InformationDavid Huggins-Daines and Alexander I Rudnicky. "A Constrained Baum-Welch Algorithm for Improved Phoneme Segmentation and Efﬁcient Training" (2006)
Available at: http://works.bepress.com/alexander_rudnicky/11/