Why don't you use large audio files with timestamps in the transcriptions?
User:
kmaclean
Date: 1/1/2010 10:40 pm
Views: 5149 Rating: -15
HTK seems to take much less time to compile
a set of smaller audio files (with no more that 10-15 utterances) rather than one large transcribed audio file (even if your transcribed
utterances have start and end time information included).