•
Speech recognition typically
produces a word- level
time-aligned annotation
•
Time alignments for other levels
of information also available
