What determines if optional groups are considered in JSGF?

[Kristoffer]

Consider the following simple JSGF. In PocketSphinx, what determines if it detects "hello" or "hello world" if I say "hello arbitrary pause world"?

public <speech> = hello [world];</speech>

Are there any parameters that control the outcome in this case? Is it possible to specify a silence threshold that means end of speech?

Or can I somehow add silence to the grammar? E.g.:
public <speech> = hello [world | sil]; // sil?</speech>

[Nickolay V. Shmyrev]

Consider the following simple JSGF. In PocketSphinx, what determines if it detects "hello" or "hello world" if I say "hello arbitrary pause world"?

There is vad module which filters silence and detects the end of the utterance.

Are there any parameters that control the outcome in this case? Is it possible to specify a silence threshold that means end of speech?

vad_threshold, vad_postspeech, vad_prespeech

Or can I somehow add silence to the grammar? E.g.:
public <speech> = hello [world | sil]; // sil?</speech>

silence is added automatically

[Nickolay V. Shmyrev]

Also, I wrote it before, it is a bad idea to impose such things in grammar. You need to process audio continuously and glue utterances for postprocessing in parsing module if you want to intelligently process them. You should not implement this logic in VAD,

You can check discussion here https://github.com/cmusphinx/pocketsphinx/issues/13#issuecomment-163304614