HMM is a probabilistic state machine in which the states of the machine are unobservable, yet the outputs are observable.
HMM with its Markov chain structure can inherently incorporate the time sequential character of the signal. By using the Gaussian mixture densities, the HMM is also expected to faithfully represent the various spectral characteristics of the HS
signal. In a recent study [22], it is found that HMM performs much better than ANN in classifying HS signals with 10
different diseases.