Why does multilingual delivery fail in players before translation quality ever does?
Different OS stacks silently substitute fonts, break line metrics, or mishandle bidirectional punctuation so overseas reviewers assume translators made mistakes. Burning captures shaping once, reducing "works on my laptop" debates. People search burn chinese subtitles mp4, arabic captions hardcoded video, rtl subtitle tips, and export subtitles for global campaigns because layout risk dominates localization schedules. Arabic and Hebrew need correct bidi isolation for numbers and parentheses. Vertical Japanese headings beside horizontal Chinese need editorial care to avoid awkward line breaks. Stacking two languages in nine-sixteen is rarely readable—ship locale-specific burns. Legal entity suffixes differ by jurisdiction—localize strings, not just word swaps. After burn you cannot let viewers toggle spoken language tracks—plan audio SKUs early. This variant sequences locale forks, native review, small-screen QA, ISO-coded filenames, then controlled distribution.
Multilingual burn variant: fork SKUs instead of cramming bilingual stacks
- Clone a project per market, localize numbers and currencies, and avoid mechanical string replace without native grammar review.
- Burn a one-minute pilot per locale, have native colleagues test on older Android with mute on, and log shaping bugs back to subtitle sources.
- Name exports with ISO language codes and revision dates, then interlink locale pages in the CMS to avoid duplicate SEO payloads.