A Hidden Markov Model (HMM) has hidden states with transition matrix A and emission distributions. The entropy rate H = lim (1/n)H(X₁,...,Xₙ) measures uncertainty in the observation sequence. It differs from the entropy of the stationary distribution.