Hidden Markov models (HMMs) provide flexible devices for modelling time series of observations that depend on underlying serially correlated states. They constitute a specific class of ...