Viterbi decoding - most probable path

Viterbi decoding to find the most probable path through given states. End state may be specified

Perform Viterbi Decoding to find the most likely path and probability of the sequence 'seq' for the system defined as follows:
p is the initial state probability matrix
A is the State transition matrix
B is the Output matrix where possible outputs are mapped to
integers 1,2,...
'end_prob' state is the final state - can be specified if required
or omitted in which case the ending state is chosen as the most probable
state
Usage:
seq=[1 1 1 2];
A=[.5 .4 .1;0 .7 .3;0 0 1];
p=[1 0 0];
B=[.8 .2;.6 .4; .1 .9];
end_state=3;
[q P]=most_likely_viterbi(seq,p,A,B,end_state);
This implementation is an improvement over similar implementations in MATLAB Central in that, the end state can be optionally specified.
Reference: This implementation is of the Algorithm described in
a speech processing course at the Univ. of Southern California, Los Angeles.
Author: Suneesh Sasikumar