Can gpt-realtime Produce Transcription for Input and Output?

[tiffiana]

Can gpt-realtime produce transcription of both input and output, so I can display the user query and model response in text in real time in addition to audio?

I suppose I can feed audio through transcription model, but I was wondering if this can be done with one model.

Thanks!

[Innovatix]

Yes, GPT Realtime can handle both. You get the audio plus text output, so you can display the user query and model response in real time. You can try it here: Realtime demo

Here’s the docs: Realtime transcription guide

https://platform.openai.com/audio/realtime