How do I caption a Shorts series in two languages in one batch?

Submit one POST /v1/video-captions job per episode per language and set language on each. A clip of at most 60 seconds costs $0.20 in Sume prices, so ten episodes in two languages is 20 jobs and $4.00.

YouTube's Help page on Shorts series describes a series as a Shorts-only playlist with seasons and episodes, with each episode at most 180 seconds (read 2026-10-07). Captions are the cheapest way to make every episode in that playlist readable with the sound off.

What the job takes

video_url is required and must be a public HTTPS video URL. style, font, language and script_text are optional. language is only a speech-to-text hint (en, es and so on); it never selects the style or the font. If you send script_text, Sume aligns your script instead of guessing the words.

The job fails with caption_no_speech on a silent clip, with next_action: use_overlay_captions; the silent micro-drama post covers that case.

A tidy batch

Burn the language you shoot in first, then the second language as a separate output file named with the episode number. Do not translate by re-running the first job; the second language needs its own script_text or its own audio.

Use design overrides to keep the same colors across the whole season; a bad value is a 400 at request time, so you do not pay for a wrong render. The price applies to videos of at most 60 seconds under the current rule, so confirm it on the captions page for longer episodes.

• Same style and design for every episode in a season.
• One job per language per episode, never one job for both.
• Keep the caption output URL next to the episode record.

Quality checks before you ship the batch

Run the first episode in each language alone before you queue the rest. Check the line length on a phone-sized preview, the color contrast on your brightest scene, and the safe margins at the bottom of the frame. A fix after one job costs $0.20; a fix after fifty costs ten dollars and a lot of waiting.

If your script is known, pass it as script_text. Speech-to-text alone can mis-hear names, and a recurring character name is exactly the word that appears in every episode. A script lets the aligner time your spelling instead of guessing.

Keep the language hint honest. language tells speech-to-text what to expect; it does not translate. A Spanish caption track needs Spanish speech or Spanish script_text, and the style or font choice is separate. The micro-drama post covers the case in which there is no speech at all.

Naming and storage

Name each output by season, episode and language, for example s01e03-es. A flat folder of fifty files called captioned.mp4 is the usual way a batch goes wrong, and it costs more to fix than the $0.20 jobs cost to run.

Store the request body next to the output. If you change the season style later, you can replay the same body with a new style and compare the two renders.