Method

Audio language detected

Spoken-language identification

What we measure and how

A 60-second clip is extracted from the middle of the programme, uploaded to a temp prefix on the ingest bucket, and submitted to AWS Transcribe with IdentifyLanguage=true. The detected primary language code is compared against the spec's expected language. For files shorter than 60 s the whole file is transcribed; below 3 s the check skips cleanly with 'audio too short for language identification'. Skips cleanly with reason 'no speech detected' for music-only or M&E tracks.

Provenance

Standards

BCP-47

Spec fields consulted

These Universal Media Delivery Profile paths resolve this check's thresholds against your spec snapshot at the moment the job was submitted.

• · assets.audio.language.expected
• · assets.audio.language.mode

The full UMDP schema: /umdp.

Related methods

• Language tag

Method identifier: audio_language_detected