Call ID: 9006da2e-1e14-4275-9e57-9562d425f1d7 Excuse my language on it, lol. Just a bit frustrated

message has been deleted

@Vapi, when I use the GPT4o realtime cluster, the system prompt is ignored. When I switch to the GPT4o cluster, the prompt is not ignored. I want to use the GPT4o Realtime Cluster. Can you fix this?

@guesty47 I understand that you're experiencing an issue where the system prompt is being ignored when using the GPT4o realtime cluster, but it works as expected with the GPT4o cluster. I'm here to help! To address this, I would need a bit more information about how you're implementing the API call and the exact behavior you're observing. Could you provide me with the details of the system prompt that's being ignored and any error messages you might be receiving? This will help me assist you better.

@User Here is the system prompt: [Identity] You are a a suicidal caller. You are incredibly depressed about your life and the direction that it's going in. [Style] - Be sad, depressed, and anxious - Be concise, as you are currently operating as a Voice Conversation. [Response Guideline] - Tone: Sound very sad, depressed and anxious [Task] 1. Tell the user about how exactly your life isn't going well and ask if they have any suggestions to help you 2. If the user suggests that you breathe in and out, do it. Otherwise, continue to complain about your life. 3. After breathing in and out, say "thank you for your help" and end the call.

@guesty47 Please ask support for more info on this.

@Vapi How do I do that?

@guesty47 To reach out for more support on this issue, you can contact the customer support team of the service provider for the GPT4o realtime cluster. They should have a support channel, such as an email address, support ticket system, or a help desk that you can use to submit this issue. Provide them with the details of the problem, including the system prompt that is being ignored and any error messages or unexpected behaviors you're encountering. If you have access to a support portal or documentation, it might also have specific instructions on how to report such issues. If you need assistance with finding the contact information or navigating the support process, please let me know, and I'll do my best to assist you further.

Hey @guesty47 I have to check about this with the team, for you reference I am sharing the eror log we recieve from openai.

logs

🔵 16:51:59:666 Received OpenAI Realtime Error. Error: { "type": "invalid_request_error", "code": null, "message": "Error committing input audio buffer: buffer too small. Expected at least 100ms of audio, but buffer only has 0.00ms of audio.", "param": null, "event_id": null }