Variable frame analysis in the ARM continuous speech recognition system
Speech Communication
Continuously variable duration hidden Markov models for automatic speech recognition
Computer Speech and Language
An efficient algorithm for parameterizing HsMM with Gaussian and Gamma distributions
Information Processing Letters
Hi-index | 0.00 |
A simple segmental hidden Markov model (HMM) is proposed which addresses some of the limitations of conventional HMM based methods. The important features of this approach are the use of an underlying semi-Markov process, in which state transitions are segment-synchronous rather than frame-synchronous and state duration is modelled explicitly, and a state segment model in which separate statistical processes are used to characterise "extra-segmental" and "intra-segmental" variability. A basic mathematical analysis of gaussian segmental HMMs is presented and model parameter reestimation equations are derived. The relationship between the new type of model and variable frame rate analysis and conventional gaussian mixture based hidden Markov models is explained.