A Hidden Markov Model is a statistical model of the distribution of "hidden" features, such as phonemes or part-of-speech tags, based on observable features, such as acoustic segments, or words. The computational models can be automatically trained from data samples, and then used to recognize the "hidden" layer, based on the statistical model derived from the training data.

(HMM) An extension of a Markov model, in which a state has a probability of emitting some output; thus, states may be "hidden."

A numeric analysis technology used frequently in continuous-speech recognition systems that recognizes speech by determining the probability of each phoneme at contiguous, small regions (frames) of the speech signal item in a string of items.

A joint statistical model for an ordered sequence of variables. The result of stochastically perturbing the variables in a Markov chain (the original variables are thus "hidden"), where the Markov chain has discrete variables which select the "state" of the HMM at each step. The perturbed values can be continuous and are the "outputs" of the HMM. A Hidden Markov Model is equivalently a coupled mixture model where the joint distribution over states is a Markov chain. Hidden Markov models are valuable in bioinformatics because they allow a search or alignment algorithm to be trained using unaligned or unweighted input sequences; and because they allow position-dependent scoring parameters such as gap penalties, thus more accurately modeling the consequences of evolutionary events on sequence families.

A probabilistic model used to align and analyze sequence datasets by generalization from a sequence profile.

a generative model that describes a probability distribution over a set of strings

a statistical model similar to a profile (Gribskov et al

A probabilistic model consisting of a number of interconnecting states. Like profiles, HMMs encode full domain alignments. They are essentially linear chains of match, delete or insert states: a match state denotes a conserved column in an alignment; an insert state allows insertions relative to match states; and delete states allow match positions to be skipped.

A method for inferring a hidden state in a system where the hidden state generates a sequence of observable events. This sequence of observable events is a Markov chain. For example, many speech recognition software programs use the sequence of phonemes in a segment of speech to infer a spoken word. In this example, the system is a person speaking, the condition of the system is a particular spoken word, and the observable events are the phonemes that make up the word.

A hidden Markov model (HMM) is a statistical model in which the system being modeled is assumed to be a Markov process with unknown parameters, and the challenge is to determine the hidden parameters from the observable parameters. The extracted model parameters can then be used to perform further analysis, for example for pattern recognition applications. A HMM can be considered as the simplest dynamic Bayesian network.