word segmentation pseudocode

take source file
take minimum silence period
take silence thresh
split whole audio into some chunks of length = minimum silence period
for every chunk
	if dBFS of a chunk is less than minimum thresh
		store the starting time of silence
	endif
	if all chunks hav beed checked
		break
	endif
endfor
combine and store all the silence ranges
find non-silence ranges using silence ranges
split the main audio into final chunks using non-silence ranges