> ## Documentation Index
> Fetch the complete documentation index at: https://kupe.in/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Model, Transcriber & TTS

> Configure the model, transcriber, and text-to-speech for your assistant.

For **HTTP API** catalog codes, language conventions, and logos, see [Model providers](/docs/assistants/model-providers).

## Model

In the **Model** section you configure how the assistant thinks and responds.

* Choose a **model** from the dropdown.
* Set **temperature** to control variability — lower values are more deterministic, higher values more creative.
* Optionally set a **welcome message** for when a call starts.
* Define a **greeting** that the agent speaks at the start of the call.

> **Use case:** Use the model settings to tune tone, creativity, and how the conversation starts.

***

## Transcriber

The **transcriber** turns the caller’s speech into text for the model.

* Pick a **transcriber** provider from the list.
* It converts speech to text in real time so the agent can respond quickly.

### Streaming

Some providers support a **Streaming** option. When enabled, transcription is **streaming**, which keeps conversations smooth and low-latency.

> **Note:** Most providers support streaming. If you use a non-streaming provider, response timing may differ.

***

## Text to speech (TTS)

**TTS** turns the assistant’s text into spoken audio.

* Choose from available **voices** and providers.
* Configure **style and accent** where the provider supports it.
* **Streaming** mode can start playback as audio is generated for faster responses.

> **Recommendation:** Use TTS streaming for more natural, responsive calls.