Kinetic Caption Video

Create captions that participate in the edit rather than behaving like fixed subtitles. Use alone on an existing HyperFrames video or alongside $social-talking-head-video and $popout-talking-head-video.

Inspect assets/example/kinetic-depth-captions.mp4 when a concrete reference for roaming and behind-presenter caption zones is useful.

Source

Require corrected word-level timestamps. Generate and group them when needed:

npx hyperframes transcribe input.mp4 -d analysis/transcript --json --model small.en
python3 scripts/group_transcript.py analysis/transcript/transcript.json --out caption-groups.json

Correct brands, products, APIs, numbers, and acronyms without changing timestamps.

Caption plan