Shopify adds Arabic, Urdu and Hebrew: what Sume captions document

Shopify's changelog says Arabic, Urdu and Hebrew are now supported in Admin and Storefronts, but Sume's caption docs do not say whether those scripts render in burned-in captions. Run one test clip before you build a localized product video batch, and keep clean copies of every clip.

The Shopify fact comes from its changelog, dated October 2, 2026. The caption facts come from Sume's video captions page.

Right-to-left scripts bring layout questions that Latin text does not: letters change shape depending on neighbors, punctuation flips, and numbers keep their own direction. A renderer that handles Latin and Hangul well is not automatically good at these, which is why a test beats an assumption.

What did Shopify announce?

The changelog entry reads "Arabic, Urdu, and Hebrew are now supported in Admin and Storefronts." The line I read does not describe layout details, which of your themes are ready, or how translated media is handled. Merchants in those markets can now run an Admin and storefront in those languages; whether their videos carry text is a separate question.

What do Sume's caption docs name?

The captions page names Latin display styles (slam, punch, tiktok-green) and a family of Hangul identities for Korean speech, with Hangul fonts. It says Korean copy on a Latin style returns a 400 rather than drawing tofu boxes. It says nothing about Arabic, Urdu or Hebrew glyphs, shaping or right-to-left layout.

A page that is silent is not a no. It means you should not promise it. Inspect and captions both take a language or language_code hint such as en or ko, and omitting it triggers automatic detection, but the pages do not list supported languages.

How do I test one clip safely?

Burn a short authored line with cues, which skips speech-to-text and draws exactly the text you give it. A 60-second-or-less caption job is a fixed $0.20 under the current estimate, so one test is cheap. Then open the result and check three things: are the letters joined correctly, does the line read right to left, and is any glyph a box?

Use a one-line test, not your full script, so a failure is easy to read.

curl -X POST https://api.sume.com/v1/video-captions \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: rtl-caption-test-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/example/clean.mp4",
    "style": "slam",
    "cues": [
      { "text": "TEST LINE", "start": 0.5, "end": 2.5 }
    ]
  }'
# Swap TEST LINE for one real sentence in the target script before you run it.

What if the render is wrong?

Then ship the clip without burned-in text and put the words where Shopify can render them: the product title, description and translated fields in the storefront. That keeps the video language-neutral, which is easier to reuse across markets. For avatar videos, remember inline captions can soft-fail and still leave a clean primary video, so the clean file is already there.

Do not assume the avatar voice speaks the language either. The avatar docs describe scripts and scenes but I found no language list, so listen to a test before you make twenty.

Also keep the product title in the clip file name or your tracker in the language your team reads, so you can find files later; storefront text can be in the shopper's language while your internal naming stays consistent.

Should I wait for Shopify's next steps?

You do not need to. A clean, caption-free clip works in any storefront language, and captions are an add-on. If you do need captions, test first and keep the source.

If you sell in these markets, ask a native reader to look at the first test clip. A render can look fine to someone who cannot read the script and still have broken letter joins, so a reviewer who can is worth more than any checklist here.