I have developed two software packages for fitting hidden Markov models of various sorts which both run under the R environment for statistical computing.
Depmix fits straightforward hidden Markov models with gaussian or multinomial responses. Moreover, it can fit mixtures of HMMs and has multi-group capabilities. The reference is: Ingmar Visser. Depmix, an R package for fitting mixtures of (latent) Markov models on multivariate mixed timeseries data. Package and manual with illustrative examples, current stable version of depmix available on CRAN.
DepmixS4 fits hidden Markov models of generalized linear models and allows fitting of covariates on both the initial state and transition probabilities using a multinomial logistic link function. There is now a paper with illustrative examples on using depmixS4 to fit hidden Markov models: Ingmar Visser & Maarten Speekenbrink (2010). depmixS4: An R-package for hidden Markov models. Journal of Statistical Software , 36(7):1–21. Online: http://www.jstatsoft.org/v36/i07/