Hi! I was wondering if it was possible to see the text transcript of the Realtime voice responses? I’d love to have a log of the responses for testing. Thank you!
You should get a response.audio_transcript.done
event when the responses have finished
Thank you! For anyone who may visit this thread in the future, you can condition on the response.audio_transcript.done event and then the response.transcript value will hold the Realtime transcript.
Is there any way we can get time stamp also into the transcript?
2 Likes
Thanks for this. very helpful. Do you know how we can also get the transcription of the human speaker’s audio (ie: the user)?
It’s not clear from the docs.
Thanks in advance.
1 Like
as far as getting user’s transcription, this event should do the work:
conversation.item.input_audio_transcription.completed
you might have to config in the session update:
"modalities": ["text", "audio"],
"input_audio_transcription": {"model": "whisper-1"},
did it work?