Skip to main content🧠 Model
In the Model section, you can configure how your assistant thinks and behaves.
- From the dropdown list, choose any available model of your choice.
- Adjust the temperature to control the variability in responses — lower values make the assistant more deterministic, while higher values make it more creative.
- Specify a welcome message to greet users when a call starts.
- Define a greeting message that the agent will speak at the start of the call exactly as written.
💡 Use Case: Use the model configuration to tailor your assistant’s tone, creativity, and initial conversation flow.
🎙️ Transcriber
The Transcriber listens to the caller’s voice and converts it into text for the model to understand.
- Select a transcriber provider from the available options.
- It transcribes spoken input into text for real-time understanding and response.
⚙️ Streaming Option
Each provider may have a Streaming Enabled toggle.
Enabling streaming allows the transcription to happen in real time, leading to faster and more natural conversations.
⚠️ Important: Most providers support streaming mode, which is preferred for smooth, low-latency conversations.
The only exception is Faster-Whisper, which operates in non-streaming mode. Be cautious when changing this setting if using Faster-Whisper.
🔊 Text to Speech (TTS)
The TTS engine converts the assistant’s text responses into natural-sounding speech.
- Choose from a wide variety of human-like voices across different providers.
- You can configure the voice style, gender, and accent depending on the provider.
- There is also a streaming mode option, allowing speech to begin playback as it’s generated — improving response time and interactivity.
💡 Recommendation: Keep TTS in streaming mode for faster, more natural responses during real-time calls.
