Files
Paul Adenot f5c06603c2 Bug 2060247 - Select the speech recognition model from the locales it declares. r=media-playback-reviewers,ai-platform-reviewers,valentinp,alwu
Each models.yaml entry now lists the locales its artifact recognizes, taken from
the artifact itself, and a request picks the first entry that recognizes it,
negotiated with LocaleService::NegotiateLanguages. That replaces the locale
prefixes and the multilingual catch-all, so a language no model recognizes no
longer reaches one that would transcribe it as another language.

Differential Revision: https://phabricator.services.mozilla.com/D322688
2026-09-09 07:56:46 +00:00
..