I’m looking at transcribing recorded meetings with OpenAI’s audio API. The easy sections are fine, but once two people start speaking at the same time, I’m not sure what to expect from the transcript.

Can it still produce useful speaker labels in overlapping audio, or does diarization tend to break down in those moments? If you’ve tested this on real meeting recordings, I’d be interested to hear whether you clean up the audio first or accept some speaker confusion.