From: Fang Liu on
Murphy's usage webpages mentions:

HMMs with mixture of Gaussians outputs
Maximum likelihood parameter estimation using EM (Baum Welch)
Let us generate nex=50 vector-valued sequences of length T=50; each vector has size O=2.

the question is

which one is the number of sequences, and which one is the length of each sequence ?

From the script of mhmm_logprob.m, I guess T is the number of sequences and nex is the length of each sequence. However, the comment in mhmm_em_demo.m gives
T = 420; % Number of vectors in a sequence
nex = 1; % Number of sequences