How to train the first mono-phone acoustic model?

I described the process to prepare the dictionary file, lexicon and use HTK for training and decoding. However, I cannot recall all the details.