Speech Segmentation Aspects of Phone Transition Acoustical Modelling
The paper presents our experiences with the phone transition acoustical models. The phone transition models were compared to the traditional context dependent phone models. We put special attention on the speech signal segmentation analysis to provide a better insight into certain segmentation effects when using the different acoustical models. Experiments with the HMM-based models were performed using the HTK toolkit, which was extended to provide proper state parameter tying for the phone transition models. All the model parameters were estimated on the GOPOLIS speech database. The annotation confusions concerning two-phone speech units are also discussed.
Unable to display preview. Download preview PDF.
- 1.Dobrišek, S. (1999). Analysis and Recognition of Phones in Speech Signals. Ph.D. Thesis in preparation, (In Slovenian). University of Ljubljana, Faculty of Electrical Engineering, Ljubljana Slovenia.Google Scholar
- 2.Dobrišek, S., Gros, J., Mihelič, F., and Pavešić, N. (1998), Recording and labelling of the GOPOLIS Slovenian speech database. Proc. 1st Int. Conf. on Language Resources & Evaluation, Vol. 2, ESCA, pp. 1089–1096.Google Scholar
- 3.Gros, J., Pavešić, N., Mihelič, F. (1997), Text-to-Speech Synthesis: A Complete System for the Slovenian Language. Jurnal of Computing and Information Technology, Vol. 5(1), pp. 11–19.Google Scholar
- 4.Young, S., Odell, J., Ollason, D., Vatchev, V., and Woodland, P. (1997), The HTK Book. Cambridge University, Entropic Cambridge Research Laboratory Ltd.Google Scholar