30-second ad sound bill: voiceover, music, render, captions

A 30-second ad with a voiceover, a music bed and burned captions costs about 45 cents on Sume at the public rates, excluding the picture. That is one TTS job, one Music generation, one Timeline render and one caption job. The sum depends on one assumption you should change: how many characters the script has.

Line by line

The script is 450 characters, which is roughly what a 30-second read holds at a normal pace. That is an assumption for this worked example, not a measurement, so replace it with your own count. TTS is $0.0475 per 1,000 characters, so 450 characters is 450 divided by 1,000 times $0.0475, or about $0.021.

The Music job is $0.125 per generation. The render is $0.10 per output minute rounded up, so a 30-second output bills as one minute. Captions are $0.20 per job for a clip of up to 60 seconds.

Where the money moves

Retakes change the total. A second voiceover take adds about two cents, but a second music generation adds 12.5 cents, so choose music takes with care. Captions are billed per job on a finished clip, so run them last, after the voiceover and the music are locked.

Ways the bill changes

A longer script is cheap: doubling it to 900 characters adds about two cents. A longer output is not much dearer either, since the render bills by the minute and a 30-second ad already rounds up to one. The big levers are the number of music takes and the number of caption passes.

Change the assumptions one at a time, and keep the plan in a spreadsheet. Run the unbilled plan call before the render, and keep every retake in the sheet. A 45-cent ad that is rerun four times is a two-dollar ad.

Use the unbilled POST /v1/timeline-1.0/plan call to see the render estimate before you commit. These are the public rates in the repository and the docs, and the docs say to confirm the live rate in GET /v1/catalog, so read it there before you quote a client.