Reach learners worldwide.

AI generation in the target language end-to-end.

• Language picker on the wizard — defaults to your browser locale.
• System prompts inject the target language as a hard directive; no English bleed-through.
• Action verbs + connectors localized — no "Explain, Identify, Apply" in Japanese output.
• Cue annotation pass adapts to the language's connectors (e.g. "donc" in French, "因此" in Chinese).

Voice library, filtered by language.

• ElevenLabs voices labelled by language — picker hides voices that don't match the course language.
• Per-scene voice override when you want a specific character.
• Plan-gated: paid tiers unlock premium voices; Free uses the OpenAI multilingual TTS.

RTL layout flip without manual work.

• Arabic, Hebrew, Persian, Urdu — the runtime flips automatically.
• Canvas-side preview honours the same flip so authors see the final shape.
• Bidirectional text handled correctly within mixed-language paragraphs.

WhisperX alignment with language auto-detect.

• Forced-alignment supports every language WhisperX does.
• Pass the course language explicitly OR let the model detect.
• Word-level timestamps stay accurate across scripts (Latin, Cyrillic, CJK, Arabic).

XLIFF export for human review.

• Export the whole course as XLIFF for your translation team.
• Re-import with their changes; the course rebuilds automatically.
• Per-string version diff so reviewers see what changed.

One source of truth, many published variants.

• Clone-and-translate — duplicate the source course, switch language, regenerate.
• Linked variants share theme + assets; copy diverges.
• Publish each variant independently to SCORM, xAPI, or LTI.