Boxes are face tracks per frame (25 fps); a box is highlighted while that speaker is talking. The timeline shows utterances per speaker and the hold/shift events between turns. Zoom in to 15 s or less to see individual words; hover a bar for its text and timing. Scroll on the timeline to zoom, drag it to pan. Keys: Space play/pause, ←/→ 5 s, J/K previous/next event, +/− zoom. These two segments are a preview; the full test set is distributed separately.