The hidden Markov model (HMM) is a powerful statistical model
that closely approximates many phenomenon found in nature,
such as human speech. While a very powerful statistical
model, a single HMM cannot easily act as
a classifier between a wide variety of signal classes.
Instead, it is best to design them specifically
for each signal type and feature type. A versatile HMM software
toolbox for MATLAB is also described.
PDF Estimation using HMMs