15th Annual Conference of the International Speech Communication Association

September 14-18, 2014

An Iterative Approach to Decision Tree Training for Context Dependent Speech Synthesis

Xiayu Chen, Yang Zhang, Mark Hasegawa-Johnson

University of Illinois at Urbana-Champaign, USA

EDHMM with decision trees is a popular model for parametric speech synthesis. Traditional training procedure constructs the decision trees after observation probability densities have been optimized with the EM algorithm, assuming the state assignment probability does not change much during tree construction. This paper proposes an iterative algorithm that removes the assumption. In the new algorithm, the decision tree construction is incorporated into the EM iteration, with a safeguard procedure that ensures convergence. Evaluation on The Boston University Radio Speech corpus shows that the proposed algorithm can achieve a significantly better optimum in the training set than the original one, and that the advantage is well generalizable to the test set.

Full Paper

Bibliographic reference.  Chen, Xiayu / Zhang, Yang / Hasegawa-Johnson, Mark (2014): "An iterative approach to decision tree training for context dependent speech synthesis", In INTERSPEECH-2014, 2327-2331.