Fits latent (hidden) Markov models on mixed categorical and continuous (time series) data, otherwise known as dependent mixture models, see Visser & Speekenbrink (2010, <DOI:10.18637/jss.v036.i07>).