Sixth International Conference on Spoken Language Processing
(ICSLP 2000)

Beijing, China
October 16-20, 2000

Data-Driven Model Construction for Continuous Speech Recognition Using Overlapping Articulatory Features

Jiping Sun, Xing Jing, Li Deng

Department of Electrical and Computer Engineering, University of Waterloo, Canada

A new, data-driven approach to deriving overlapping articulatory-feature based HMMs for speech recognition is presented in this paper. This approach uses speech data from University of Wisconsin's Microbeam X-ray Speech Production Database. Regression tree models were created for constructing HMMs. Use of actual articulatory data improves upon our previous rule-based feature overlapping system. The regression trees allow construction of the HMM topology for an arbitrary utterance given its phonetic transcription and some prosodic information. Experimental results in ASR show preliminary success of this approach.

Full Paper

Bibliographic reference.  Sun, Jiping / Jing, Xing / Deng, Li (2000): "Data-driven model construction for continuous speech recognition using overlapping articulatory features", In ICSLP-2000, vol.1, 437-440.