How does vad work?

The online documentation isn’t completely clear on how it works:

Voice Activity Detector. Attempts to trim silence and quiet background sounds from the ends of (fairly high resolution i.e. 16-bit, 44−48kHz) recordings of speech.

…

Options: 
Default values are shown in parenthesis. 
−t num (7)

The measurement level used to trigger activity detection. This might need to be changed depending on the noise level, signal level and other charactistics of the input audio.

What is a “measurement level”?  What does the default number 7 represent?

I’m trying to use vad to trim silence from a spoken phrase like “guess what”.  Using vad to trim the front of the audio clip eliminates the first half of the word “guess” and I’m trying to figure out how to fix that.

Thanks…

Damon