Anyone experiencing big response time 4sec+ after human first start to reply to AI 1st msg? I'm noticing that on 70%+ of my cold calls, using make and vonage phone nb https://cdn.discordapp.com/attachments/1332286268805087272/1332286269044031488/response_time.png?ex=6794b3c7&is=67936247&hm=0e4015ef9d97f5497a0a2b27815beceffdb72b5704f064669170355cf9bc375d&

@Jetsoccer Game I'm sorry to hear that you're experiencing longer response times. It's important to ensure that your voice applications remain responsive. There could be several factors contributing to the delay, such as network latency, API performance, or the complexity of the voice application logic. To help you better, could you provide more details about your setup? For instance, are you using any specific VAPI features that might be affecting performance? Additionally, have you noticed any patterns in the calls that could help isolate the issue, such as time of day, call duration, or particular voice commands that precede the delay? Understanding these details can help us troubleshoot the problem more effectively. If you have already gathered this information and need further assistance, please let me know, and we can look into the issue more deeply.

@Jetsoccer Game It seems like you're experiencing consistent delays with your MAKE+VAPI calls. Here are a few leads on what to check: 1. **Network Latency**: Verify the network latency from your system to the API endpoints. High latency can contribute to delays. 2. **API Performance**: Check the API performance metrics and logs to see if there are any reported slowdowns or issues. 3. **Voice Application Logic**: Review the logic of your voice application to ensure there are no inefficient processes that could be causing the delay. 4. **Resource Utilization**: Monitor the server or service hosting your voice application for CPU, memory, and bandwidth usage. 5. **Third-party Services**: If your application relies on third-party services or APIs, check their status and response times as they could be a bottleneck. 6. **Call Volume**: High call volumes can sometimes slow down response times. Check if the delays correlate with peak usage times. 7. **VAPI Configuration**: Review your VAPI configuration, especially any settings that might affect call setup and response times, such as

silenceTimeoutSeconds

,

responseDelaySeconds

, and

llmRequestDelaySeconds

. Regarding the mention

<@&1231325148079194222>

, it appears to be a user or role mention from a platform like Discord or Slack, which typically notifies the user or members of the role. However, without context, it's hard to determine the exact meaning or status it implies. If this is related to a support ticket or a specific issue within a communication platform, you might want to check directly within that platform or contact their support for more information. If you need further assistance with the delays you're experiencing, please provide additional details or consider reaching out to the support team for MAKE+VAPI.

@Vapi what does this value "Phone call provider clos..." mean for 'Ended reason'

@Jetsoccer Game The value "Phone call provider clos..." for 'Ended reason' seems to be truncated, but it likely indicates that the call was ended because the phone call provider closed the connection. This could happen for a variety of reasons, such as: - The call was completed successfully, and the provider terminated the connection as expected. - There was an issue on the provider's side, such as a technical glitch or maintenance activity, leading to an unexpected disconnection. - The call was ended by the user or recipient, prompting the provider to close the connection. If you're seeing this as a common 'Ended reason' and it's not expected behavior (e.g., calls are being dropped unexpectedly), you may want to reach out to your phone call provider's support team to investigate the issue further. They can provide more detailed insights into why the connection was closed and help resolve any underlying problems.

so it means the user just hang up because probable not interested ?

@Jetsoccer Game Apologies for any confusion. The 'Ended reason' value "Phone call provider clos..." could indeed suggest that the call was terminated by the phone call provider, which might happen if the user hung up the call, possibly indicating they were not interested. However, without the full text, it's difficult to be certain of the exact reason. If the calls are ending prematurely and this is a concern, it would be wise to investigate further. You can check call logs for patterns, reach out to the phone call provider for more detailed explanations, or review the interaction flow to ensure there's nothing in the voice application that might lead to a call being ended unexpectedly. If you're seeing this reason frequently and it's not clear why users might be hanging up, consider examining the user experience to ensure that the voice prompts are engaging and that there's no confusion on the part of the caller that could lead them to end the call.

@User any idea? this happens on hundreds of outbound calls via make.com

@Jetsoccer Game I will be addressing this matter after a period of time, as I am currently focused on some scheduled tasks. Apologies for the delay.

sure, booked a slot with you since we're doing hundreds of outbound calls via make but sometimes response time is under 2sec and often 3+sec

@Jetsoccer Game You have set smartEndpointingEnabled to true, which only works for English transcriptions, so set it to false. Secondly, you are using Deepgram, and to capture the words better or capture all, including single words, set Deepgram's endpointing to 300ms. Because of this, your single words were not captured properly, along with others. Lastly, set waitSeconds of startSpeakingPlan to 0\.8 seconds, because Deepgram and this startSpeakingPlan (our models including Deepgram's) detect if the user has stopped speaking within the ample period of time required.

Hi @Shubham Bajaj cannot set the ms on deepgram

@Jetsoccer Game You have to update it using UPDATE /assistant/:id API.

curl -X PATCH https://api.vapi.ai/assistant/assistant-id \
     -H "Authorization: Bearer token-here" \
     -H "Content-Type: application/json" \
     -d '{
  "transcriber": {
    "provider": "deepgram",
    "endpointing": 300
  }
}'

@Shubham Bajaj https://cdn.discordapp.com/attachments/1332286268805087272/1333009879971004497/image.png?ex=679755b1&is=67960431&hm=43df7caf348b5c06f7a0c785d0147db57d0fb14af8b34a0fa89d8b374a5ae502&

@Jetsoccer Game you need to pass the authentication token, assistantId correctly I suggesting using API client for this purpose https://docs.vapi.ai/api-reference/assistants/update and your missing auth-token as visible from the error response.

@Shubham Bajaj actually you can send CURL via VAPI api, very cool

@Jetsoccer Game just replace the required values such as private-key or token, assistantId https://discord.com/channels/1211482211119796234/1332286268805087272/1332669604400726108

@Shubham Bajaj it's done