A 59-second voice-over Short with captions: $0.34 in Sume jobs

A 59-second voice-over Short with burned-in captions costs $0.342 in Sume jobs when the picture is already made: about $0.042 for the voice, $0.20 for captions and $0.10 for a one-minute render. Aim for 59 seconds because the standalone caption estimate in the docs covers videos of at most 60 seconds, and the render rounds up to whole minutes, so 61 seconds doubles it to $0.20.

How long a script to write

Sume TTS bills by character, so the script length sets the voice cost. At an assumed 15 characters a second, 59 seconds is 885 characters. That is 0.885 x $0.0475 = $0.0420375. Measure your own voice and replace 15; a slow narrator needs fewer characters.

The ledger

Prices are from the Sume catalog and the video captions docs. The clip generation that supplies the picture is not included.

Watch the 60-second edge

The Video captions docs say each accepted caption job reserves $0.20 for videos of a maximum of 60 seconds under the current fixed estimate, and tell you to check GET /v1/catalog and OpenAPI for the live price. The catalog I read lists no separate caption price, so confirm before you scale. If your video runs longer, trim it, or plan for the live rate.

The render is simpler: Timeline bills $0.10 per output minute, rounded up, with a one minute minimum. A 60.0 second video is one minute, and 60.1 seconds is two, $0.20.

Order of operations

Generate the voice first and measure its length. Then build the picture to that length so nothing is cut after the fact. Render with the voice as the audio spine, caption the result, and use the speech check in the captions docs: a clip with no audible speech fails with caption_no_speech and wants authored cues. If your script has unusual spellings, send script_text so the burned words match your copy while the speech timings stay the source of time. See the video captions docs for style names and the Korean rules.

If the video runs long

A 90 second version with the same method has about 1,350 characters of script, which is $0.064125 in TTS, and two render minutes at $0.20. The caption line is the one to confirm, since the fixed estimate is stated for up to 60 seconds. Total before captions: $0.264125. Check the live catalog price for captions, add it, and you have the number.

Trimming a long script

If your script reads at 66 seconds, you have two honest choices: cut about 105 characters or raise the speed. Cutting is better. A speed of 1.12 would bring 66 seconds to 59 but is audibly faster than a normal read. Cutting also lowers the TTS cost, by 105 characters x $0.0000475 = $0.005, a small but welcome saving.