This paper addresses the problem of parameter estimation in the case of hidden data. The aim is to discuss two general iterative parameter estimation methods "Expectation-Maximization" (EM) and "Iterative Conditional Estimation" (ICE) in the context of the classical Hidden Markov Models (HMMs) and in the context of the recent Triplet Markov Models (TMMs). A very general method of TMMs identification based on ICE and copulas is also specified