Localize a Claude Motion explainer: four caption languages for $0.80

To ship a Claude Motion explainer in four languages, submit four Sume caption jobs on the same clip, each with its own cues. Each job is $0.20 for a video of 60 seconds or less, so English, Spanish, French and Korean cost 4 x $0.20 = $0.80. Authored cues skip speech-to-text, so the silent clip and the translation both work.

A caveat on scope: this captions an existing video, it does not re-animate the on-screen text. Anthropic's announcement says Motion animates text as code and lets you change any word, number or timing, so for fully localized pictures, regenerate the animation in each language and caption only where you want an extra layer.

One job per language

Every request carries the same video_url and a different cues array of text, start and end in seconds. You can send only one of script_text, words, cues and segments in a request. Use a separate Idempotency-Key for each language so a retry of the Spanish job does not replay the English one.

The Korean exception

Sume's Latin styles do not take Hangul. Sending Hangul text on a Latin style such as slam returns a 400, and if you do not send a style the text of the captions picks one: slam for Latin and black-outline for Korean. So leave style unset for Korean or choose a Hangul style, and for korean-ad also send language: "ko".

Reusing the first job

If you only want a different look for the same words, a restyle with source_caption_id avoids a second speech-to-text pass. With authored cues there is no speech-to-text to save, and the docs say a restyle is still a render, so it bills $0.20 as well.

The clip must be at a public HTTPS URL, and signed or private URLs are rejected. Host it once and reuse the same URL for all four requests.

Keeping cues in step with the animation

Cue timing is the part that can drift. Spanish and French run longer than English for the same sentence, so a cue that fits 2.5 seconds in English may need 3 seconds or a shorter phrase in the translation. Because cues carry explicit start and end times, you set each one yourself instead of trusting a transcript. Keep a table of cues per language next to the animation script, and when someone edits a number in the animation, update all four cue sets before you submit.

Submit the English job first and look at the result. If the placement or style is wrong, fix it once and reuse the settings for the other three, so you pay for one trial job and not four.

Budget for a weekly series

If a weekly explainer goes out in four languages, that is $0.80 a week, or 52 x $0.80 = $41.60 a year, plus whatever the video itself costs. Add languages one at a time: each additional language is a flat $0.20 for a clip under 60 seconds. A clip over 60 seconds is outside the price the docs give, so keep localized explainers under a minute or check the live catalog rate.