In electrical engineering, statistical computing and bioinformatics, the Baum–Welch algorithm is a special case of the EM algorithm used to find the unknown parameters of a hidden Markov model (HMM). It makes use of the forward-backward algorithm to compute the statistics for the expectation step.