THREE SCENES / 24 SECONDS — ORIGINAL TEACHING SOURCES
Created 26 October 2026. Fictional memories, original table illustrations and disclosed Microsoft Zira synthetic narration. No family testimonial or AlbumMap product recording.

Use in your editor: import the three PNGs and matching WAVs. Place each image for eight seconds, beginning at 0, 8 and 16 seconds. Put each voice take at its scene start without stretching it. The remainder of each scene is quiet. The MP4 example has no music. Import the VTT where supported, or use the transcript to create checked captions.

Rebuild locally: with Python 3 and FFmpeg available, run `python build.py` from this extracted folder. This creates scene MP4s, concat.txt and voiceover-three-scenes-rebuilt.mp4 inside this folder. It never uploads files. Existing outputs with these names are overwritten; source PNG/WAV files are not modified. Extract into a fresh folder.

Replacing the speech: record your own voice or use an appropriately permitted voice. Keep each take under eight seconds or change the scene duration and captions together. The supplied caption chunks divide measured utterances evenly; adjust them to your actual pauses.

Review: confirm all three scenes, the final 24-second duration, understandable speech, and captions on the player you will use. The illustration is a placeholder, not an invented family photograph.
