@Vapi @Vapi Support Bot I'm working on an assistant that needs to handle various text elements like numbers in a more natural, human-like way. For example: - "123" should be spoken as "one hundred twenty-three" - "2024" could be "twenty twenty-four" or "two thousand twenty-four" ## The Question Does Vapi provide any built-in settings or features for text normalization that converts these elements into their spoken equivalents? I'm looking for something that would automatically handle this conversion without having to manually define every possible case. I am using 11labs. Different TTS engines seem to handle this differently, and I'm wondering if there's a Vapi-specific way to control this behavior.

Yes, Vapi has built-in text normalization for numbers. To control how numbers are spoken (e.g., "123" → "one hundred twenty-three"), use the

numberToDigitsCutoff

setting in your assistant's configuration. For example:

json
"voice": {
  "chunkPlan": {
    "formatPlan": {
      "enabled": true,
      "numberToDigitsCutoff": 2025
    }
  }
}

Numbers below the cutoff are spelled out; numbers above are read as digits. See the [Vapi voice formatting plan documentation](https://docs.vapi.ai/assistants/voice-formatting-plan) for details. Source: - [Voice formatting plan documentation](https://docs.vapi.ai/assistants/voice-formatting-plan)

Hi, thanks for reaching out.