HyperFrames: multiple sequential elements on the same data-track-index cause audio bleed at clip boundaries — a previous clip's audio leaks into the next clip's time window. Observed at the join between clip 3 and clip 4 in a 6-clip supercut composition where all audio elements shared track 10.
Put each element on its own data-track-index (e.g. tracks 11, 12, 13, ...) instead of sharing a single audio track. The HyperFrames runtime manages audio playback per-track, and same-track sequential audio clips don't get cleanly silenced at their end boundary. Using separate tracks eliminates the overlap entirely. This is consistent with the tracks-and-clips reference ('tracks are temporal, not visual') but the docs don't warn that same-track sequential audio can bleed.