Principle:Openai Openai node Realtime Conversation Interaction
| Knowledge Sources | |
|---|---|
| Domains | Realtime, Conversation |
| Last Updated | 2026-02-15 00:00 GMT |
Overview
A principle for conducting multi-modal conversations over the Realtime API by sending client events that create conversation items, append audio, and trigger model responses.
Description
Realtime Conversation Interaction covers the main interaction loop of a Realtime session. The client sends typed events to: add text or audio messages to the conversation (conversation.item.create), append audio buffer data (input_audio_buffer.append), commit audio input (input_audio_buffer.commit), and explicitly request model responses (response.create).
The interaction supports both text-only and audio-based conversations, with the model generating text and/or audio responses depending on the session's configured modalities.
Usage
Use this principle during the active conversation phase of a Realtime session. Send events to interact with the model after session configuration.
Theoretical Basis
Conversation interaction follows an Event-Driven Message Passing pattern:
// Text conversation:
rt.send({ type: 'conversation.item.create', item: {
type: 'message', role: 'user',
content: [{ type: 'input_text', text: 'Hello!' }]
}})
rt.send({ type: 'response.create' })
// Audio conversation:
rt.send({ type: 'input_audio_buffer.append', audio: base64AudioChunk })
rt.send({ type: 'input_audio_buffer.commit' })
rt.send({ type: 'response.create' })