Fact · Speech recognition
Speaker diarization is not supported in Gemini live transcription sessions, and Google points to its non-streaming audio transcription endpoint for speaker labels.
Checked against its source on October 17, 2026
In the source's own words
“Speaker diarization is not supported in live streaming sessions. For speaker diarization, use the non-streaming Audio transcription endpoint.”
Live transcription with Gemini Live API, Google, Google AI for Developers, 2026, Limitations.
Open the original- What it measures
- Google's live transcription guide lists this among the limitations of the streaming model. The same guide lists word-level timestamps as unsupported in the live session as well.
- What it does not show
- This applies to the live streaming model. The batch model's page lists speaker diarization as a feature, so a recording that needs speaker labels should go through the non-streaming route.