What we listen for in a sermon

This file is a placeholder so the blog renderer has something to render. It will be replaced by the real seed posts.

The loudest thirty seconds is rarely the truest thirty seconds

When we first tried picking moments from loudness alone, a calm preacher's highest peaks turned out to be an explanation of a map. Volume tells you where the room reacted, not where the meaning landed.

The transcript carries most of the signal: a rhetorical question, a phrase repeated three times, a direct instruction, a sentence that finishes a thought. The audio adds the rest, mainly pauses, pace and the sound of a room responding.

What that means for your reels

1. A quiet invitation can be chosen over a shout.
2. Cuts land where a sentence breathes, not on a fixed timer.
3. Worship music is separated from the sermon before anything is picked, so the reel is never a slice of the band.

You still read the captions before anything goes out. That part is not automated, and it should not be.