The assistant I built is taking a very long time to respond when someone answers the phone and says "hello?" (~10 seconds). The weird thing is that sometimes it works fine and sometimes there is this huge delay. The unfortunate part is that I am presenting this to a company that wants to replace a 30 person call center and I had this come up both times I tried to demo it. The first time was two days ago when there seemed to be a problem with the platform, and the second time was this morning. Something that may be worth noting is that the agent is part of a conference call. In the conference call, the conference call first calls the ai, then it calls the customer, and then it calls a sales agent. I do not think this is the problem considering this used to work well and did not start giving me problems till two days ago. If you could help me with this, I would appreciate it. The assistant id is: 64cd1eab-5a05-43f9-80ee-6c2370746146 Here are some examples, but it happened a lot more: - 1bea349d-2f20-44b3-b64f-f8e8894f6c2b - aa08330c-d5e1-4556-9eb8-47d5b8453960 - 040ed21c-1e72-4f90-9ace-b2b8e5f32c10

@Abe I'm sorry to hear about the issues you've been experiencing with the response times during your demonstrations. Delays can be quite frustrating, especially in a professional setting where you're showcasing the capabilities of your assistant. To address this, I would recommend checking a few things: 1. Network Latency: Ensure that the network connection is stable and robust, as network issues can cause delays in the assistant's responses. 2. Server Load: High server load can also lead to increased response times. If possible, check the server status or reach out to the platform support to see if there were any known issues during the times you experienced delays. 3. Conference Call Complexity: While you mentioned that the conference call setup worked well before, it's worth revisiting the configuration to ensure that there haven't been any changes that could affect the assistant's performance. 4. Assistant Configuration: Review the assistant's configuration to ensure that there are no new or changed settings that could be contributing to the delay. 5. Platform Updates: Check if there have been any recent updates to the platform that might have affected the assistant's performance. 6. Logs and Metrics: Review the logs and metrics for the assistant around the times of the delays to identify any patterns or issues. If after checking these areas the problem persists, it may be beneficial to reach out to the platform's support team with the assistant ID and the example call IDs you've provided. They might be able to offer more insight into the issue or identify if there's a problem on their end. Since you've provided the assistant ID and specific examples, I can also initiate a support ticket for you to get this issue looked into by the technical team. Would you like me to go ahead and do that for you?

I tried again just now and the delay was not as bad. However, last night it was also working well and then this morning there was a big delay again. Could the big delay be due to high server load that occurs during the day?

@Abe this is from 11labs try cartesia.

I cannot do that because I have to use a cloned voice for this project. Can you provide me with details on what is wrong with eleven labs and why it sometimes works? Also, in the section where I can hear the recording and see the audio waves, I see a green lines right before the Ai talks. What do those green lines represent?

- the issue happens because of streaming and traffic cannot be controlled - these are visual representations of the different speakers

When I looked at the logs, it looks like the model request sent by vapi is taking several seconds to be sent instead of being sent right after someone says "hello?". Can you tell me why the model request is taking so long, please?

it depends upon endpointing time based and punctuation based.

Here are some examples, that show the model request sent by vapi is taking several seconds: 1bea349d-2f20-44b3-b64f-f8e8894f6c2b aa08330c-d5e1-4556-9eb8-47d5b8453960 040ed21c-1e72-4f90-9ace-b2b8e5f32c10

set optimizeStreamingLatency to 2