Each word in the normalised text needs a pronunciation. Most words will be found in the dictionary, but for the remainder we must predict pronunciation from spelling.
in Dan Jurafsky and James H. Martin “Speech and language processing: an introduction to natural language processing, computational linguistics, and speech recognition”, 2014, Pearson Prentice Hall, Upper Saddle River, N.J., Second edition, ISBN 0135041961
University of Edinburgh online access
Scanned copy of chapter 8 (University of Edinburgh students only)
