https://vapi.ai/ logo
Join Discord
Powered by
@Shubham Bajaj can you point me to the
# general-english
h
@Shubham Bajaj can you point me to the documentation for the new vapi voicemail detection? the UI is broken and wont save, I can find google, openAI and Twilio detection plans on swagger, but nothing on Vapis https://cdn.discordapp.com/attachments/1211484481848873020/1365423073729183744/image.png?ex=680d40d5&is=680bef55&hm=da40e0f35f95488876085b534da1226fe2a75a3791d38656ffb6fa57103c77a3&
s
twilio
curl -X PUT \ https://api.vapi.ai/assistant/YOUR_ASSISTANT_ID \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "voicemailDetection": { "provider": "twilio", "enabled": true, "machineDetectionTimeout": 30, "machineDetectionSpeechThreshold": 2400, "machineDetectionSpeechEndThreshold": 1000, "machineDetectionSilenceTimeout": 5000, "voicemailDetectionTypes": [ "machine_end_beep", "machine_end_silence", "machine_end_other" ] } }' ref: https://docs.vapi.ai/calls/voicemail-detection#1-using-twilios-voicemail-detection
google
curl -X PUT \ https://api.vapi.ai/assistant/YOUR_ASSISTANT_ID \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "voicemailDetection": { "provider": "google", "voicemailExpectedDurationSeconds": 25 } }'
openai
curl -X PUT \ https://api.vapi.ai/assistant/YOUR_ASSISTANT_ID \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "voicemailDetection": { "provider": "openai", "voicemailExpectedDurationSeconds": 25 } }'
h
the rest of that /\
s
curl -X PUT \ https://api.vapi.ai/assistant/YOUR_ASSISTANT_ID \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "voicemailDetection": null }'
@Ham what else do you require?
h
your UI is showing a recomended "vapi" voicemail detection. Your UI will not save any changes to this setting The support bot suggested changing one value and leaving the rest the same. Where is the full schema for adding the recomended vapi voicemail detection?
the only thing I can find is the voicemail tool
s
can you change your org channel to weekly from daily.
h
it is weekly
the bot in there gave a few lines but where is the full schema?
I would rather just ad it with patch assistant than troubleshoot the UI
s
vapi
Copy code
export interface VapiVoicemailDetection {
  /** This is the provider to use for voicemail detection. */
  provider: 'vapi';
VoicemailDetectionBackoffPlan;

  beepMaxAwaitSeconds?: number;

  enableTwilioBeepDetection?: boolean;
}
curl -X PUT \ https://api.vapi.ai/assistant/YOUR_ASSISTANT_ID \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "voicemailDetection": { "provider": "vapi", "backoffPlan": { "startAtSeconds": 5, "frequencySeconds": 5, "maxRetries": 6 }, "beepMaxAwaitSeconds": 30, "enableTwilioBeepDetection": true } }'
@Ham let me know if you require further help.
h
Copy code
{
  "message": [
    "voicemailDetection.an unknown value was passed to the validate function"
  ],
  "error": "Bad Request",
  "statusCode": 400
}
same error no matter what i put in tried matching keys, tried different values
a
even am facing this same issue
s
This is coming for which voicemail type? could you record a loom video of it so i can share with the team.
h
what is voicemail type? Vapi? on the dashboard the voicemail detections are listed : Off Vapi (recommended) Google OpenAI Twilio You can not publish any changes unless you select Twilio We are trying to use your (recommended) voicemail detection Vapi I have tried patching my assistant with the schema you provided and I get the same error as the dashboard no matter what I put in for values
there is no record of a vapi voicemail detection anywhere in the support docs, the schema you provided does not match the dashboard entries, or what the AI bot said in support thread
this is the dashboard error
and this is the request error
Copy code
{
  "message": [
    "voicemailDetection.an unknown value was passed to the validate function"
  ],
  "error": "Bad Request",
  "statusCode": 400
}
its the same error
j
First, thank you Shubham for all the help that you provide. You do a great job. I attempted to change the voicemail detection to Vapi and started having errors as well. Not able to publish and the assistant is now lagging for multiple seconds. I belive there may have been an update that caused a bug.
h
s
@Ham @Jay @Arjun for now can you guys use twilio/google/openai instead for voicemail?
h
the only one that will load is twilio
I can get openai to load but only with these values, The values that the dashboard is trying to load throw a should not exist error "voicemailDetection": { "provider": "openai", "voicemailExpectedDurationSeconds": 15 },
swagger just crashes out of memory with all these voicemail detection errors when you open create call
s
@Ham @Jay @Arjun can you update your assistant's voicemail config using the dashboard now?
h
sucessfully updated thank you!
actually it says its published but if you refresh its showing no changes
patching assistant works, dashboard isnt throwing errors anymore but it does not save the publish changes "voicemailDetection": { "provider": "vapi", "backoffPlan": { "maxRetries": 6, "startAtSeconds": 5, "frequencySeconds": 5 }, "beepMaxAwaitSeconds": 2 },
j
I am having the same problem. Also, now when I test an outbound call the asssistant says the first line then goes right into the voicemail sequence and hangs up.
h
using the twilio detection yesterday 1 out of 4 or 5 calls it would play first message and voicemail message if detecting a voicemail I managed to patch in vapi voicemail detection today I havent tested it yet
j
Wish I could test it. Even though my telco provider Telnyx says i have no rate limit set, vapi says I have reached my rate limit 13 calls in smh.
Still have long awkward pauses of 7 seconds or longer after each human response. Last test pause was 16 seconds. If i could get this fixed I could finally actually use this. @Shubham Bajaj, @User I have changed start speaking plan to 0 wait seconds, enabled vapi smart endpointing, have letancy at 850ms, Stop speaking plan is set to number of words (0), voice seconds (0.2), and back off seconds (1). I do not see the response delay, llm request delay, or pipeline configuration in my dashboard. Transcriber is set to deepgram nova 3 (although I have also tried nova 2). I have tried using azure voices but that didnt change anything. I am trying these calls from the outbound number on the phone numbers tab, as well as through a make.com scenario. Im not sure what else I can try. The only place that the assistant does not have these delays is when I speak to it through the talk to assistant button at the top of the screen.
I was finally able to test it. That seemed to stop the assistant from going directly into the voicemail script and hanging up mid sentence. Thanks.
s
@Jay can you create a new support ticket for endpoint issues? Please share the call IDs and ping me and King.
4 Views
PreviousNext
Powered by