RTMS sends the same text Zoom is generating for the meeting’s automated captions, so the transcript language follows the meeting’s spoken “Caption language” (and Zoom’s auto-detection, which can take ~10 seconds before it locks in).
The host/admin must allow automated captions and can limit which languages are available; Zoom explicitly says “the availability of languages is determined by the host’s automated captioning settings” in both host caption management and participant caption language selection. If the host restricts who can enable captions, participants have to use the built-in “Request captions” flow and the host approves it.
If captions are available, a participant can set “Caption language” from their Zoom client, but it impacts all participants. An RTMS app can only suggest a “Default Source Language for Transcripts,” and Zoom notes auto-detection will override it after ~30 seconds (RTMS changelog); for translated output, use Zoom’s translated captions or translate RTMS text yourself.