Imagine we have a markov chain but we don't know the current state but instead we have another set of observed variables and we know the probablity of each of them happening depending on the statue of the hidden markov chain. That is called a hidden markov model (HMM)
It calculates the most probable sequence of events for the observed output