Here is my current order of operations:
(A) I open connection # standard convo set up here
(B) {type: session.update, …} # I set the session details / instructions here
After audio input I hit
(C) {type: conversation.item.create, item: {“type”: “message”, “role”: “user”, “content”: {“type”: “input_audio”, “audio”: $base64data}}
(D) {“type”: response.create}
This works for me in terms of starting the conversation and getting audio from the real time api to render to my user
When I replace (C) and (D) with
(C) {“type”: “input_audio_buffer.append”, “audio”: $base64data}
(D) {“type”: “input_audio_buffer.commit”}
I get the error stated above
Error committing input audio buffer: the buffer is empty.
I’d like to use input_audio_buffer to take advantage of server side VAD and more to make the experience truly interactive.