Does "out-of-band" responses degrade GPT-4o Realtime?

API
1

Hi there!

I’ve been experimenting with the demo GitHub - openai/openai-realtime-agents: This is a simple demonstration of more advanced, agentic patterns built on top of the Realtime API.
and I was wondering if creating responses outside the default conversation (for analysis of the user input like the example below) has the same effect of extending the system prompt of GPT-4o Realtime, giving too many responsabilities to it and thus degrading its quality.

For example, let’s say I want to analyse n different aspects of the user audio input.

 prompt = """
Analyze the conversation so far. If it is related to support, output
"support". If it is related to sales, output "sales".
"""

event = {
    "type": "response.create",
    "response": {
        # Setting to "none" indicates the response is out of band,
        # and will not be added to the default conversation
        "conversation": "none",

        # Set metadata to help identify responses sent back from the model
        "metadata": { "topic": "classification" },

        # Set any other available response fields
        "modalities": [ "text" ],
        "instructions": prompt,
    },
}

ws.send(json.dumps(event))