Track Language Metadata
Track language metadata identifies the language of an audio or subtitle track for display and selection.
An Emby audio selector can offer two tracks called English even though one is the main film and the other is commentary. Language metadata answers which language a track represents; it does not describe the track's role.
| Value encountered in a report | Interpretation |
|---|---|
| eng | English in ISO 639-2 |
| deu or ger | German terminology/bibliographic ISO 639-2 codes |
| pt-BR | Portuguese with the Brazil region subtag in BCP 47 |
| und | Undetermined language |
A machine-readable tag helps selection software compare a component with the viewer's preferences. A human-readable title supplies extra description such as “Commentary” or “Stereo.” Both can appear together, and neither needs to be copied into the other.
Where regional detail fits
A BCP 47 tag can identify language and optional attributes such as script or region. The hyphen separates subtags; pt-BR is not a new unrelated language code. This richer form is useful where a distinction between regional variants matters to the viewer.
Containers use different language fields. Matroska has a legacy field and a LanguageIETF field for BCP 47. When LanguageIETF is present, it takes precedence under the specification. Updating only the legacy code can thus leave the effective language unchanged for a reader using the newer field.
That gives a concrete explanation for a misleading edit: the track title now says “English,” its old field says eng, but its LanguageIETF field still identifies German. The visible title and the machine-readable effective tag describe different things.
Comparing two selection results
For each relevant audio or subtitle track, collect its effective language, title, and role/selection flags. Then compare them with the user's audio and subtitle preferences. This isolates a language mismatch from a commentary-role mismatch.
Subtitles describe the language of their displayed text, which can differ from the soundtrack. The library's original-language metadata describes the work as a whole; it need not match every available translation.
A metadata edit can correct the description without decoding the sound. It does not translate the recorded dialogue. Reading the fields after the edit establishes that the intended description, including any higher-precedence field, is now the one the software receives.