If you have excessive silences at the start or end of many of your recordings, you might want to endpoint them. Only try this if your forced alignment does not give good results.
This part is optional. If you find that the forced alignment step does not give very good alignments, then try endpointing your utterances to remove excessive amounts of silence, then re-run the forced alignment. Here’s a script for end pointing.
for FILE in *.wav do sox $FILE ../silenceTrimmed/$FILE vad -t 6 -s 0.1 -p 0.1 reverse vad -t 4 -s 0.1 -p 0.3 reverse done
You’ll need to have two versions of your wav directory (before and after end pointing) and make sure that the ‘after endpointing’ versions are used in all subsequent steps, starting with the forced alignment.