I'm building a Turkish voice assistant on Vapi. Certain words like "BMW" and "Range Rover" are being mispronounced by the TTS. I tried the following: ElevenLabs: Attempted to use pronunciation dictionaries via

pronunciationDictionaryLocators

in the voice config, but it doesn't seem to work for Turkish — the custom pronunciations are not applied. Cartesia: Cartesia's own API supports

pronunciation_dict_id

, but there is no way to pass this parameter through Vapi's voice config for Cartesia. Is there a supported way to define custom pronunciations for Turkish language assistants on Vapi? Any workaround or roadmap info would be appreciated.

Hi there, Thank you for your message. Our team is currently out of the office. We operate Monday through Friday, from 9:00 AM to 8:00 PM Pacific Standard Time (PST). We’ll get back to you as soon as possible during our normal business hours. If your message is urgent, please mark it accordingly or include “URGENT” in the subject line, and we’ll do our best to respond promptly. Warm regards, Vapi Customer Support Team

Hi, I’ve handled this exact issue before. Vapi doesn’t properly support custom pronunciation for Turkish yet, so those configs won’t apply. The clean workaround is rewriting words like “BMW” or “Range Rover” into phonetic Turkish before they hit TTS. I can set that up so it runs automatically and sounds natural. Happy to help. @Deadroom

Hello, first of all, thank you very much for your response. If you’re suggesting a solution like typing the pronunciations of brands such as “BMW” into the prompt, I’m already entering those values as custom variables. That’s why I can’t enter such a command into the prompt. I need a solution that will automatically pronounce these words the way I specify when it detects them. Do you know of such a solution? @Adam

Yes — there is a way

Hi, To address your issue with number normalization in Turkish while using the TTS system, disabling voice formatting completely should prevent the system from applying English-centric normalization. Here's how you can do that:

{
  "voice": {
    "chunkPlan": {
      "formatPlan": {
        "enabled": false
      }
    }
  }
}

Applying this change will stop the formatting from converting numbers and expressions to English before they reach the TTS engine. If this approach still causes issues, consider utilizing the replacements feature to manually adjust specific expressions. For example:

{
  "voice": {
    "chunkPlan": {
      "formatPlan": {
        "replacements": [
          { "type": "exact", "key": "three plus one", "value": "üç artı bir" }
          // Add more replacements as needed
        ]
      }
    }
  }
}

This method allows for precise control over how specific phrases are pronounced until a more comprehensive solution is available. For further customization, consult the [Voice Formatting Plan documentation](https://docs.vapi.ai/assistants/voice-formatting-plan).