Scribe v2 Realtime already supports automatic language detection. To receive the detected language code, enable it at session start by setting include_language_detection=true in the WebSocket handshake; committed_transcript_with_timestamps events will then include language_code. (elevenlabs.io)
ElevenLabs’ launch post lists “Automatic language detection: Speak in any language, switch language mid conversation” as a key feature of Scribe v2 Realtime. (elevenlabs.io)
The public API docs reference the model_id as scribe_v2_realtime; there is no public “v2.1” model identifier documented. Use scribe_v2_realtime with the flag above. (elevenlabs.io)
The Speech-to-Text capability page also highlights smart language detection support. (elevenlabs.io)
tawdry carbon
tardy goblet