deep-jade•2w ago

Voice chunk plan not working effectively

For Assistant-id 4ecbe09b-0f8d-4d81-bf27-da39cdc6dd83 I've voice configuration as:

    "voice": {
        "model": "eleven_turbo_v2_5",
        "speed": 1.16,
        "style": 0.07,
        "voiceId": "yM93hbw8Qtvdma2wCnJG",
        "provider": "11labs",
        "stability": 0.6,
        "similarityBoost": 0.75
    }

    "voice": {
        "model": "eleven_turbo_v2_5",
        "speed": 1.16,
        "style": 0.07,
        "voiceId": "yM93hbw8Qtvdma2wCnJG",
        "provider": "11labs",
        "stability": 0.6,
        "similarityBoost": 0.75
    }

I'm assuming that with this configuration default chunk plan should be enabled. So I'm expecting digits like 1111 to be transformed when sent to elevenlabs. However, in log for call-id 6861480c-fb64-4304-9c40-e4c947f4641c I'm seeing:

  {
    "id": "log-92",
    "level": 30,
    "time": 1755830961324,
    "timestamp": 1755830961324000000,
    "severityText": "INFO",
    "severityNumber": 9,
    "body": "Voice input",
    "attributes": {
      "category": "voice",
      "callId": "6861480c-fb64-4304-9c40-e4c947f4641c",
      "orgId": "ca47ffe4-7368-44b7-860e-863a82ab347f",
      "input": "Just to confirm, your street address is 1111 Second Street."
    }
  }

  {
    "id": "log-92",
    "level": 30,
    "time": 1755830961324,
    "timestamp": 1755830961324000000,
    "severityText": "INFO",
    "severityNumber": 9,
    "body": "Voice input",
    "attributes": {
      "category": "voice",
      "callId": "6861480c-fb64-4304-9c40-e4c947f4641c",
      "orgId": "ca47ffe4-7368-44b7-860e-863a82ab347f",
      "input": "Just to confirm, your street address is 1111 Second Street."
    }
  }

And the pronunciation is not good. How can I fix it?

Voice formatting plan | Vapi

Format LLM output for natural-sounding speech

3 Replies

sensitive-blue•2w ago

Having similar issue

deep-jadeOP•7d ago

@Shubham Bajaj @Vapi Team can you please suggest? This is blocking production deployment.

Shubham Bajaj•4d ago

Please try using this snippet for the voice object too update:

{ "voice": { "model": "eleven_turbo_v2_5", "speed": 1.16, "style": 0.07, "voiceId": "yM93hbw8Qtvdma2wCnJG", "provider": "11labs", "stability": 0.6, "similarityBoost": 0.75, "chunkPlan": { "enabled": true, "formatPlan": { "enabled": true, "numberToDigitsCutoff": 999 } } } } Let us know If you continue to experience issues

Voice chunk plan not working effectively

Did you find this page helpful?