Can someone tell me what would be the best settings for realistic model?

Hi there! 👋 I don't have specific information about what would be the "best" settings in an absolute sense, as optimal settings often depend on your specific use case. However, I can definitely share the available settings you can configure for a realistic voice model in Vapi! For a realistic voice model, here are the key settings you can configure: ## Model Selection Select

gpt-4o-realtime-preview-2024-12-17

as your model for a realistic experience. Note that when using this model, only OpenAI voices can be selected as the voice used within the speech-to-speech model. [According to the documentation](https://docs.vapi.ai/openai-realtime). ## Voice Quality Settings You can adjust these parameters to enhance realism: - `stability`: Try setting to 0.5 to address fluctuation in tonality - `similarityBoost`: A value of 0.75 is suggested - `style`: Setting this to 0.0 is generally advisable for stability (higher values may cause unpredictability) ## Chunking Plan Settings The

chunkPlan.enabled

setting controls how text is processed:

json
{
  "voice": {
    "chunkPlan": {
      "enabled": true,
      "minCharacters": 30, // Adjust between 1-80
      "punctuationBoundaries": [".", "!", "?", ";"] // Customize based on your needs
    }
  }
}

## Adding Voice Realism in Prompts To make voices sound more natural, you can incorporate: - **Stuttering**: "I-I-I don't know" - **Hesitations**: "I was, uh, thinking about it" - **Pauses**: "I… I don't know how to say this" - **Emotional emphasis**: Using capital letters or exclamation marks Is there anything specific about these settings you'd like me to explain in more detail? Sources: - [docs.vapi.ai/openai-realtime](https://docs.vapi.ai/openai-realtime) - [docs.vapi.ai/prompting-guide](https://docs.vapi.ai/prompting-guide) - [discord.com/channels/1211482211119796234/1211483291191083018/threads/1339794428688076871](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1339794428688076871)

Hey, you can use 11labs for the voice provider, deepgram for the transcribers and open AI GPT 4o for the LLM.