Add a voice-over

Narration is three jobs — writing the script, recording the voice, timing it to the picture — and your agent can do all three.

What you need

• A composition that needs narrating (or a brief for one).
• A connected agent with access to a text-to-speech tool.

What to say

Write a 20-second voice-over for this video — confident, plain-spoken — generate it, and time the scenes to the read.

Or, with your own script:

Record this script as the voice-over and stretch each scene to match its line.

What happens

The agent drafts (or takes) the script, generates the audio with a text-to-speech tool it has access to, and uploads it as an ordinary audio track. Then it works the timing both directions — nudging scene durations to breathe with the read, or tightening the script where a scene shouldn’t wait. Because it wrote the script, captions come free: ask and it lays word-timed captions from the same text.

Make it yours

The narration is a normal audio track — trim it, duck music under it, or re-generate a single line by asking. Captions ride the transcript, so a re-recorded line updates its captions with it (see caption a talking head).