Conclusions
Hidden Markov models for sequence labeling

For the past twenty years, Hidden Markov Models have been an essential tool for processing, exploring, classifying, labeling and clustering sequential data and signals of all kinds, from audio signals (speech, music) to gestures, handwriting and human-computer interaction sequences.

Hidden Markov models provide a simple and effective framework from which the designer can easily build models adapted to a specific problem and particular data, as evidenced by the multitude of variants and extensions of these models, even if their use requires particular attention and a certain expertise.