Title :
Cross validation and Minimum Generation Error for improved model clustering in HMM-based TTS
Author :
Feng-Long Xie ; Yi-Jian Wu ; Soong, Frank K.
Author_Institution :
Microsoft Res. Asia, Beijing, China
Abstract :
In HMM-based speech synthesis, context-dependent hidden Markov model (HMM) is widely used for its capability to synthesize highly intelligible and fairly smooth speech. However, to train HMMs of all possible contexts well is difficult, or even impossible, due to the intrinsic, insufficient training data coverage problem. As a result, thus trained models may over fit and their capability in predicting any unseen context in test is highly restricted. Recently cross-validation (CV) has been explored and applied to the decision tree-based clustering with the Maximum-Likelihood (ML) criterion and showed improved robustness in TTS synthesis. In this paper we generalize CV to decision tree clustering but with a different, Minimum Generation Error (MGE), criterion. Experimental results show that the generalization to MGE results in better TTS synthesis performance than that of the baseline systems.
Keywords :
decision trees; hidden Markov models; maximum likelihood estimation; pattern clustering; speech synthesis; HMM-based TTS; HMM-based speech synthesis; MGE; context-dependent hidden Markov model; cross validation; decision tree-based clustering; maximum-likelihood criterion; minimum generation error; Context; Decision trees; Hidden Markov models; Speech; Speech synthesis; Training; Training data; HMM-based synthesis; context clustering; cross validation; minimum generation error;
Conference_Titel :
Chinese Spoken Language Processing (ISCSLP), 2012 8th International Symposium on
Conference_Location :
Kowloon
Print_ISBN :
978-1-4673-2506-6
Electronic_ISBN :
978-1-4673-2505-9
DOI :
10.1109/ISCSLP.2012.6423459
