AI Video Captions

LearnHouse can automatically generate closed captions (subtitles) for hosted video activities and translate them into any languages you choose. Captions are transcribed from the video’s audio with AI, so you don’t have to write or time them by hand.

Enabling captions

You can set captions up in two places — the panel is the same in both:

• When uploading a video — the AI Closed Captions panel appears in the new-video dialog, so captions start generating as soon as the upload finishes.
• Later, from the editor — edit an existing hosted-video activity to add, change, or regenerate captions.

To configure them:

1. Open the new-video dialog (or edit a hosted-video activity).
2. In the AI Closed Captions panel, turn on Generate with AI.
3. Pick the spoken language of the video, or leave it on Auto-detect.
4. Choose the caption languages you want. You can:
   • select any of the platform’s built-in languages, and/or
   • add a custom language by entering its code (e.g. sw for Swahili, es-419 for Latin-American Spanish) and a display name.
5. Click Generate captions.

Generation runs in the background. Each language shows a status — queued → processing → ready — and captions appear in the player’s subtitles/CC menu as soon as they’re ready. You can close the editor and come back later; you don’t need to keep it open.

How it works

1. The audio is extracted from your video.
2. The AI transcribes it into timed WebVTT subtitles in the original language.
3. Each requested language is translated from that transcript, preserving the timings.
4. The finished tracks are stored alongside the video and served to the player.

Long videos are transcribed in segments and stitched back together, so a two-hour lecture works the same as a two-minute clip.

Playback

Once at least one language is ready, learners get a CC / subtitles button in the video player and can switch languages from its menu. Captions work on both the adaptive (HLS) and standard playback paths.

AI usage & credits

Caption generation uses your organization’s AI credits, the same pool used by the AI chat and course tools.

• The cost scales with the video’s length and the number of languages you request (roughly one credit per ~10 minutes of audio, plus one per translated language).
• If your organization has AI disabled or has run out of credits, you’ll be told when you try to generate — nothing is charged in that case.
• If the source language is among your chosen languages, it is reused directly and not re-translated (saving a credit).

Regenerating or disabling

• Re-open the panel and change the languages, then Generate captions again to update.
• Turn off Generate with AI and save to disable captions for the activity.

Requirements (self-hosted)

Captions are available automatically wherever these are in place — no extra feature flag:

• A configured AI provider (Gemini by default — see the AI settings) with the AI feature enabled for the organization.
• ffmpeg available to the API (already included in the official image).
• Redis (used to queue and process caption jobs).