Skip to main content
Every video uploaded to ShortKit is analyzed for spoken language automatically. The result is stored as customMetadata.language on the content item using a BCP 47 language code. No configuration is required.

How it works

  1. You upload a video — no special parameters needed.
  2. Once the video reaches ready status, language detection runs in the background.
  3. Within a few minutes, customMetadata.language is set on the content item.
Language detection runs asynchronously after the video reaches ready status. The language key will not be present immediately when uploadStatus first transitions to ready — expect it within a few minutes.

The language field

customMetadata.language contains a BCP 47 language code (e.g. "en", "es", "ja"). It appears on the content object alongside any other custom metadata you set at upload time.
"und" is returned when no spoken language is detected — for example, music-only or ambient audio content.

Language codes

Detection covers a broad range of languages. The table below lists the most commonly returned codes, but others may appear as valid BCP 47 codes.

Filtering the feed by language

Use FeedFilter.metadata to serve a feed that contains only content in a specific language. The SDK applies the filter server-side on every request — initial load, pagination, and pull-to-refresh.
You can combine a language filter with other filters. For example, to show only Spanish-language sports content:
See Configuration for the full FeedFilter reference.